1. Introduction

Impressed by cultural diversity in beautification practices, Darwin (1874) concluded that humans lacked universal, biologically based standards of beauty. However, evidence that perceptions of attractiveness are similar across cultures (Langlois et al., 2000) and emerge early in development (Rubenstein et al., 1999, Slater et al., 1998) challenge this view. Evolutionary psychologists propose that perceptions of attractiveness are species-wide, sexually selected adaptations for finding good mates (Fink & Penton-Voak, 2002, Rhodes & Zebrowitz, 2002, Thornhill & Gangestad, 1999). If attractive traits and preferences for them are under sexual selection, then those traits should be associated with individual variation in mating success. The present study was designed to test this hypothesis.

In most animals, differential parental investment by males and females leads to different mating strategies. A greater investment by females (gestation, lactation, etc.) means that their reproductive potential is lower than that of males (Clutton-Brock & Parker, 1992, Trivers, 1972, Williams, 1975). Males, with a minimum investment in sperm, can increase their reproductive success by obtaining fertilizations with several females, although they may also benefit from providing parental care in the context of long-term relationships. Females have less to gain from mating with several males (Arnold & Duvall, 1994, Bateman, 1948), and it is more important to secure male parental investment for reproductive success (Buss & Schmitt, 1993, Clutton-Brock, 1991, Daly & Wilson, 1983, Symons, 1979, Trivers, 1972). Such investment can be provided either by one (monogamy) or a series (serial monogamy) of long-term mates. Therefore, if attractive individuals have a reproductive advantage over their peers, then we would expect attractive males to have more short-term mating success, and attractive females to have more long-term mating success than their peers. The lifetime reproductive success of both sexes can also be enhanced by becoming sexually active as soon as the individual is fertile. Therefore, we would expect attractive traits to be negatively correlated with age of first sex in both males and females.

For species that form pair bonds, both males and females can gain additional reproductive success through extra-pair copulations (EPCs). Males can always enhance their reproductive success through seeking additional mates, and in some human societies, men will seek EPCs given the opportunity (e.g., Marlowe, 1999). Females can also benefit from EPCs. In nonhuman animals, such as birds, females who are paired with low-quality males have been shown to enhance their reproductive success by seeking EPCs with males of higher quality than their long-term partner (Hasselquist et al., 1996, Kempenaers et al., 1992). Male EPC activity is likely to be constrained primarily by females' willingness to participate. Therefore, for males, we would expect attractive traits to be positively correlated with EPCs (Gangestad & Thornhill, 1997, Møller & Thornhill, 1998). In contrast, for females, we might expect attractive traits to be negatively correlated with EPCs if attractive females are able to attract males of high quality as their long-term partners.

Physical attractiveness is important in a potential mate (Buss & Schmitt, 1993), but so are other traits, and the precise role of attractiveness in mating behavior remains unclear. Surprisingly, few studies have investigated the relationship between attractiveness and reproductive behavior, and these have generally used poor measures of attractiveness (either self-reported attractiveness or attractiveness rated by only a few observers; for a review, see Wiederman & Hurst, 1998). A few studies have examined the relationship between the attractive trait of body symmetry and sexual behavior (Gangestad & Thornhill, 1997, Thornhill & Gangestad, 1994). Low levels of body symmetry (high asymmetry) appear to be associated with low self-reported lifetime number of sexual partners (controlling for age) in men and women, late self-reported age of first sex in men (Thornhill & Gangestad, 1994), and low numbers of EPCs in men (Gangestad & Thornhill, 1997). These results have been presented as evidence that sexual selection favors individuals of high genetic quality, as indicated by low levels of fluctuating asymmetry (FA). However, FA was not formally isolated from other forms of asymmetry, which do not reflect mate quality, rendering the conclusions equivocal (Tomkins & Simmons, 2003). No studies have examined the relationship between sexual behavior and facial symmetry or other attractive face or body traits.

Here, we examined whether attractive individuals have higher mating success than their peers, as expected if traits that contribute to attractiveness are under sexual selection. Specifically, we investigated whether face and body attractiveness correlate with sexual behaviors that would be associated with reproductive success in the absence of contraception. To determine which attractive traits are responsible for any such correlations, we examined correlations between sexual behaviors and three components of attractiveness: averageness, symmetry, and sexual dimorphism (masculinity in males, femininity in females). For faces, these traits are attractive across cultures (for reviews, see Fink & Penton-Voak, 2002, Rhodes & Zebrowitz, 2002, Thornhill & Gangestad, 1999) and may reflect developmental stability and health (Møller & Swaddle, 1997, Rhodes et al., 2001, Thornhill & Møller, 1997). Evidence that facial sexual dimorphism is attractive is clearer for femininity than for masculinity (for a review, see Zebrowitz & Rhodes, 2002), but an association with health has only been found for masculinity (Rhodes, Chan, Zebrowitz, & Simmons, 2003). Body symmetry and sexual dimorphism are both attractive (Gangestad et al., 2001, Gangestad & Thornhill, 1997, Singh, 1993, Singh, 1995, Thornhill & Gangestad, 1994, Tovée et al., 1999, Tovée et al., 2000). No studies have yet assessed the attractiveness of body averageness, but it seems likely to be attractive given the curvilinear association between body mass index and attractiveness (for female bodies), which can be recast as a linear association with body averageness (Tovée et al., 1999). Finally, to validate our approach, we measured height, which is known to correlate with reproductive success (greater for tall men and short women) in contemporary societies (Mueller & Mazur, 2001, Nettle, 2002a, Nettle, 2002b, Pawlowski et al., 2000).

Attractiveness and its components were assessed by ratings because measurements are generally too simplistic, relying on a limited number of landmark points that provide partial information about facial structure but no information about fat deposits, skin texture, or color (e.g., Grammer & Thornhill, 1994, Koehler et al., 2004, Simmons et al., 2004). Raters are sensitive to subtle differences in facial averageness, symmetry, and sexual dimorphism, and ratings of averageness and symmetry vary systematically with physical manipulations of these traits (Rhodes et al., 2000, Rhodes et al., 1998, Rhodes & Tremewan, 1996). Facial symmetry ratings are also sensitive to differences in facial FA but not directional asymmetry (Simmons et al., 2004), and ratings of sexual dimorphism are sensitive to variation in sexually dimorphic traits, such as the chin and jaw (Koehler et al., 2004). We assumed that raters would also be sensitive to variation in these traits in bodies. We also measured body FA, to facilitate comparison with previous studies showing that this is associated with sexual behavior (Gangestad & Thornhill, 1997, Thornhill & Gangestad, 1994).

2. Method

3. Results

3.1. Sex differences in reported sexual behavior

Males and females did not differ in age (for males, M=23.5, S.D.=6.1, range=18–47; for females, M=22.9, S.D.=5.6, range=17–51; Mann–Whitney U test, normal approximation Z=−0.99, p=.324). Nor did the proportions of ethnicities differ between the male and female groups [135 (153) male (female) Caucasian and 38 (43) other; χ 2(1)=3.06, P=.081]. It was therefore not necessary to consider these variables as covariates in our between-sex analyses.

Compared with females, males reported significantly more short-term and had more liberal attitudes to sexual relationships (Table 1). Both males and females behaved in accordance with their attitudes to sexual relationships, with significant correlations (controlling for age) between CAS and all behavior variables, except age of first sex in females. Correlations were negative, indicating that individuals with conservative attitudes (high CAS scores) had fewer sexual partners.

Table 1. Descriptive statistics, t tests, and Mann–Whitney U tests for male/female differences in sexual behavior variables, and Kendall partial rank-order correlations of sexual behavior variables with composite attitude score, controlling for age

	Sexual behavior variable	Sex	n	Median	Range	Mean (S.D.)	Mann–Whitney U (Z)	p	Kendall τ	p
	Age at first sex	F	158	17	13–26	17.3 (2.2)	0.46	.644	.05	ns
		M	144	17	12–27	17.6 (2.6)			.17*	.005
	Number of sexual partners	F	196	2	0–100	5.8 (11.6)	2.40	.017	−.28*	.001
		M	166	3	0–280	12.7 (29.6)			−.38*	.001
	Short-term partners	F	192	0	0–95	2.5 (7.6)	3.18*	.002	−.31*	.001
		M	160	1	0–100	7.7 (19.2)			−.38*	.001
	EPCs	F	194	0	0–15	0.5 (1.4)	1.54	.125	−.13*	.005
		M	165	0	0–50	1.5 (5.2)			−.27*	.001
	Long-term partners	F	192	1	0–7	1.0 (1.1)	−2.03	.043	−.20*	.001
		M	161	1	0–6	0.8 (0.9)			−.14*	.005
	Composite attitude score	F	196	26	4–36	25.3 (7.4)	−5.95*	.001	n/a	n/a
		M	165	20	4–36	20.4 (7.7)			n/a	n/a

* Survive Bonferroni correction for multiple comparisons (n=6 comparisons per test); critical p=.0083.

3.2. Attractiveness and reported sexual behavior

Face and body attractiveness were significantly correlated for females (Kendall's τ=.258, p<.0001, n=192) but not for males (Kendall's τ=.032, p=.55; n=164). Therefore, we kept face and body attractiveness distinct in the analyses. Table 2 shows the descriptive statistics for attractiveness and the other appearance variables, and Table 3 shows the Kendall partial correlations of face and body attractiveness with the sexual behavior variables, controlling for age. Males with attractive faces and bodies had more short-term partners than their peers, and males with attractive bodies became sexually active earlier than their peers. Females with more attractive faces had more long-term relationships and became sexually active earlier than their peers. The median number of long-term relationships for both sexes was one (Table 1).

Table 2. Descriptive statistics for attractiveness and its components, for faces and bodies

	Appearance variables		Males				Females
			n	Mean (median)	S.D.	Range	n	Mean (median)	S.D.	Range
	Faces	Rated attractiveness	166	2.8 (2.8)	0.9	1.2–5.2	196	2.9 (2.9)	0.9	1.3–6.3
		Rated sexual dimorphism	166	4.5 (4.5)	0.8	2.1–6.0	196	3.8 (3.8)	0.9	1.7–6.3
		Rated averageness	166	4.1 (4.2)	0.7	2.1–5.7	196	4.2 (4.2)	0.7	1.7–5.5
		Rated symmetry	166	4.0 (4.0)	0.7	2.5–5.8	196	4.1 (4.1)	0.8	1.6–6.3
	Bodies	Rated attractiveness	164	3.5 (3.4)	0.9	1.3–5.5	192	3.6 (3.6)	0.8	1.1–5.3
		Rated sexual dimorphism	164	4.4 (4.4)	0.9	2.1–6.1	192	4.1 (4.3)	0.9	1.6–6.1
		Rated averageness	164	4.3 (4.4)	0.8	1.1–5.7	192	4.2 (4.3)	0.7	1.4–5.4
		Rated symmetry	164	4.5 (4.4)	0.6	2.8–6.0	192	4.4 (4.5)	0.7	2.7–6.3
		Measured asymmetry (FA)	166	0.02 (0.01)	0.01	0.00–0.06	195	0.02 (0.02)	0.01	0.00–0.05

Rated averageness is reverse scored from rated distinctiveness. Rated sexual dimorphism refers to ratings of masculinity for male faces and bodies, and ratings of femininity for female faces and bodies. Body photographs were missing for six participants.

Table 3. Kendall partial rank-order correlations of attractiveness with sexual behavior variables, controlling for age

			Males			Females
			Kendall τ	p	n	Kendall τ	p	n
	Faces	Age at first sex	−.0595	ns	144	−.2184*	.001	158
		Number of sexual partners	.1174	.025	166	.1024	.01	196
		Short-term partners	.1764*	.001	160	.0197	ns	192
		EPCs (cheating)	.0614	ns	165	.0038	ns	194
		Long-term partners	−.0323	ns	161	.1225*	.005	192
	Bodies	Age at first sex	−.1463*	.005	142	.0048	ns	157
		Number of sexual partners	.1782*	.001	164	.0129	ns	192
		Short-term partners	.2070*	.001	159	−.0712	ns	188
		EPC (cheating)	.0791	ns	163	−.0809	ns	190
		Long-term partners	.0035	ns	160	.0350	ns	188

* Survive Bonferroni correction for multiple comparisons (n=10 correlations on same sample); critical p=.005.

3.3. Components of attractiveness and sexual behavior

To examine which components of attractiveness affect sexual behavior, we correlated each of rated sexual dimorphism (masculinity in males, femininity in females), averageness, and symmetry of the face and body with each sexual behavior variable. These traits were all attractive, as expected (Table 4). We did not correlate height with body attractiveness because our photographs were taken so that the body filled the frame, thereby removing visual cues to height.

Table 4. Kendall partial rank-order correlations of attractiveness with components of attractiveness, for faces and bodies, controlling for age

			Males			Females
			Kendall τ	n	P	Kendall τ	n	P
	Faces	Rated sexual dimorphism	.256*	166	.001	.602*	196	.001
		Rated averageness	.189*	166	.001	.234*	196	.001
		Rated symmetry	.183*	166	.001	.292*	196	.001
	Bodies	Rated sexual dimorphism	.573*	164	.001	.450*	192	.001
		Rated averageness	.536*	164	.001	.150*	192	.001
		Rated symmetry	.221*	164	.001	.194*	192	.001
		Measured asymmetry (body FA)	−.043	164	ns	−.087	191	.05

* Survive Bonferroni correction for multiple comparisons (n=8 correlations on same sample); critical p=.0063.

Sexual dimorphism was consistently associated with sexual behaviors likely to enhance reproductive success in both males and females (Table 5). Masculine males (bodies, with similar trends for faces) had more sexual partners, particularly short-term partners, than their less masculine peers, whereas feminine females (faces) had more long-term partners. Feminine females (faces) also became sexually active earlier than their peers. Taller males had more short-term partners than shorter males (Table 5), and taller females also had more short-term partners than shorter females.

Table 5. Kendall partial rank-order correlations of rated sexual dimorphism and height with sexual behavior variables, controlling for age

		Appearance	Sexual behavior	Male			Female
				Kendall τ	n	P	Kendall τ	n	P
	Faces	Rated sexual dimorphism	Age of first sex	−.130	144	.01	−.221*	158	.001
			Sexual partners	.137	166	.005	.117	196	.01
			Short-term partners	.112	161	.025	.038	192	ns
			EPCs (cheating)	.071	165	ns	−.059	194	ns
			Long-term partners	.050	162	ns	.152*ns	192	.001
	Bodies	Rated sexual dimorphism	Age of first sex	−.076	142	ns	.059	157	ns
			Sexual partners	.179*	164	.001	−.010	192	ns
			Short-term partners	.167*	160	.001	−.090	188	ns
			EPCs (cheating)	.099	163	.05	−.088	190	ns
			Long-term partners	.035	161	ns	−.003	188	ns
		Height	Age of first sex	−.127	143	.025	.014	158	ns
			Sexual partners	.115	165	.025	.095	196	.025
			Short-term partners	.163*	160	.001	.167*	192	.001
			EPCs (cheating)	−.014	164	ns	−.043	194	ns
			Long-term partners	−.066	161	ns	−.031	192	ns

* Survive Bonferroni correction for multiple comparisons (n=15 correlations on same sample); critical p=.0033.

Average males (face and body) had more short-term partners than their less average peers (Table 6). Males with average bodies also had more EPCs than their less average peers. For females, averageness did not correlate significantly with any of the sexual behavior variables. Symmetry showed little association with sexual behavior (Table 7). The only significant effect was that women with symmetric faces became sexually active earlier than their peers.

Table 6. Kendall partial rank-order correlations of rated averageness (reverse scored distinctiveness ratings) with sexual behavior variables, controlling for age

		Appearance	Sexual behavior	Male			Female
				Kendall τ	n	P	Kendall τ	n	P
	Faces	Rated averageness	Age of first sex	−.047	144	ns	−.111	158	.025
			Sexual partners	.099	166	.05	.030	196	ns
			Short-term partners	.156*	161	.005	.020	192	ns
			EPCs (cheating)	.049	165	ns	.017	194	ns
			Long-term partners	.021	162	ns	.036	192	ns
	Bodies	Rated averageness	Age of first sex	−.082	142	ns	−.027	157	ns
			Sexual partners	.116	164	.025	.004	192	ns
			Short-term partners	.156*	160	.005	.040	188	ns
			EPCs (cheating)	.150*	163	.005	.023	190	ns
			Long-term partners	−.014	161	ns	−.017	188

* Survive Bonferroni correction for multiple comparisons (n=10 correlations on same sample); critical p=.005.

Table 7. Kendall partial rank-order correlations of symmetry variables with sexual behavior variables, controlling for age

		Appearance	Sexual behavior	Male			Female
				Kendall τ	n	P	Kendall τ	n	P
	Faces	Rated symmetry	Age of first sex	.097	144	.05	−.170*	158	.001
			Sexual partners	−.020	166	ns	−.048	196	ns
			Short-term partners	−.020	161	ns	−.100	192	.025
			EPCs (cheating)	.027	165	ns	−.046	194	ns
			Long-term partners	.068	162	ns	.009	192	ns
	Bodies	Rated symmetry	Age of first sex	.007	142	ns	−.081	157	ns
			Sexual partners	.030	164	ns	.127	192	.005
			Short-term partners	.041	160	ns	.065	188	ns
			EPCs (cheating)	.038	163	ns	−.028	190	ns
			Long-term partners	−.092	161	.05	.125	188	.01
		Measured asymmetry (body FA)	Age of first sex	−.031	144	ns	−.017	157	ns
			Sexual partners	.060	166	ns	−.068	195	ns
			Short-term partners	.105	161	.025	−.019	191	ns
			EPCs (cheating)	−.006	165	ns	−.042	193	ns
			Long-term partners	−.052	162	ns	−.080	191	.05

* Survive Bonferroni correction for multiple comparisons (n=15 correlations on same sample); critical p=.0033.

4. Discussion

Attractive men and women were more successful in implementing their “preferred” mating strategies according to parental investment theory. Men with attractive faces and bodies enjoyed significantly more short-term mating success than their peers, with no cost in their long-term mating success, whereas women with attractive faces had more long-term mating success than their peers. Attractive men (bodies) and women (faces) also became sexually active earlier than their peers, which would enhance reproductive success for both sexes in the absence of contraception. These results support the assumption underlying much current research on attractiveness, that attractive traits are important in mate choice and may be sexually selected.

Attractiveness enhanced mating success for both men and women, suggesting that both sexes choose partners with attractive traits. This finding challenges claims that male market value is principally determined by earning potential and relationship commitment (e.g., Pawlowski & Dunbar, 1999). Attractiveness of the face and body contributed independently to male mating success because they were uncorrelated, but both affected success. Only facial attractiveness contributed to female mating success, which is surprising if attractive female bodies advertise fertility, as widely believed (e.g., Singh, 1993). However, males may have been able to assess this information from the face because face and body attractiveness were correlated in females (see also Thornhill & Grammer, 1999). It is not clear why the face and body should signal similar information in females but not in males. Future research is needed to determine the relative contributions of face and body attractiveness to overall attractiveness in each sex.

We also examined the association between sexual behaviors and three components of attractiveness: sexual dimorphism, averageness, and symmetry. We confirmed that these traits were attractive in both faces and bodies, including masculinity in male faces, which is not always found (cf., Rhodes et al., 2003). A novel finding was that averageness was attractive in bodies, just as it is in faces, where it is associated with health during development (Rhodes et al., 2001). The only trait that did not correlate with attractiveness was measured body FA. The fact that rated, but not measured, body (a)symmetry correlated with attractiveness may reflect the very limited set of bilateral traits typically used to measure body FA.

Sexual dimorphism showed the strongest associations with sexual behaviors. It was associated with short-term mating success in males and long-term mating success in females, mirroring the results for attractiveness. Specifically, men with masculine bodies had more short-term partners (similar trend for men with masculine faces) and more sexual partners overall, and women with more feminine faces had more long-term relationships and became sexually active earlier than their peers.Hughes and Gallup (2003) have reported similar associations between body sexual dimorphism and sexual behaviors in an undergraduate sample (60 males, 56 females). They measured dimorphic body shape using the shoulder-to-hip ratio for males and waist-to-hip ratio for females. Like us, they found an association with the number of sexual partners for men, although they also found an association for women, which we did not (correlation close to zero). Unlike the present study, they did not distinguish between short- and long-term partners. They also reported significant associations with age of first sex, which we found for females with feminine faces, but not bodies, and which we did not find for masculine males. However, their male correlation would not survive Bonferroni correction for multiple correlations, which may resolve the discrepancy. Finally, they also found an association with EPCs (both cheating on one's own partner and poaching the partners of others). We examined cheating EPCs and found no such correlation, but their EPC cheating correlations would not survive Bonferroni correction. Despite these discrepancies, both studies indicate that sexual dimorphism affects mating success.

The success of masculine men in attracting short-term mates may be consistent with findings that masculine faces are not consistently attractive to women, becoming relatively more attractive at the fertile phase of the menstrual cycle, and particularly for short-term relationships (Penton-Voak et al., 1999). Given that short-term mating success was associated more strongly with masculinity of the body than the face, we suggest that ovulating females might show an even stronger shift in their preference for men with masculine bodies than for men with masculine faces. Future research is needed to test this prediction.

Previous studies have shown that taller men have higher mating success (e.g., Pawlowski et al., 2000). In our study, taller men had more short-term relationships than their peers, consistent with these results and providing some validation of our methods. Interestingly, taller women also reported more short-term partners than their peers. This could reflect a nonpreferred “masculine” mating pattern, or poorer retention of mates by tall women, who generally have lower reproductive success (Nettle, 2002b).

Averageness in men showed similar relationships to sexual behaviors as found for masculinity, with more average men (faces and bodies) having greater short-term mating success than their peers. Men with more average bodies also had more EPCs. For females, there were no significant associations between averageness and sexual behaviors. Therefore, averageness may be more important for male than female mating success.

The only significant association between rated symmetry and sexual behavior was that women with symmetric faces became sexually active earlier than their peers. We found no significant correlation between body FA and mating success (or attractiveness), failing to replicate previous reports that body FA is associated with low lifetime numbers of sexual partners in men and women and the late onset of sexual activity in men (Thornhill & Gangestad, 1994). This failure to replicate cannot be due to insufficient power because we had a much larger sample than Thornhill and Gangestad did. Our analyses were conservative in two respects. First, we guarded against Type I errors using Bonferroni correction. Nevertheless, there were few individually significant correlations between sexual behaviors and FA. Second, we only included, in our composite measure of body FA, traits for which measurements of asymmetry were distributed as FAs. For direct comparison with Thornhill and Gangestad (1994), we repeated our analyses using a composite measure of asymmetry using all body traits measured, irrespective of whether asymmetry values showed the statistical properties of FA. This less conservative analysis likewise failed to replicate the result of Thornhill and Gangestad (1994) (Rhodes, Simmons, & Peters, unpublished). Gangestad and Thornhill (1997) reported a small but significant effect of body FA on EPCs in men, which we also failed to replicate here (with a similar sized sample). However, an odd feature of their data was that the number of EPCs did not increase with age, which raises doubts about the accuracy of these self-reported EPCs.

As typically found in sexual survey data, men reported more sexual partners and more liberal attitudes to sex than women did (Oliver & Hyde, 1993, Wiederman, 1997). Men also reported more short-term, but not more long-term, partners than women. These discrepancies in the numbers of partners claimed by men and women are paradoxical because in a closed population with an equal sex ratio, the number of partners must match for men and women. The reasons behind these discrepancies have been discussed widely (for a review see Wiederman, 1997). The important point here is that they do not affect associations between appearance and sexual behaviors (calculated separately for males and females), which were the focus of this study.

This research has some limitations. Like much research on attractiveness, it was based on a primarily university sample. The youthfulness of such a sample necessarily limits variance in mating success. Another limitation is that appearance was rated from single, front-view photographs of each participant (except for height and body FA). However, despite these limitations, we found clear associations between attractiveness (and its components) and mating success.

Is there any evidence that the associations between behavior and morphological traits found here could actually influence reproductive success in modern populations? Recent studies of male height suggest that they might. Taller men are generally more attractive than those of average height (Feingold, 1982, Gillis & Avis, 1980, Hensley, 1994, Jackson & Ervin, 1992, Shepperd & Strathman, 1989). Consistent with this finding, we found a general tendency for tall men to have more sexual partners, particularly short-term partners, than short men. Studies from three modern human populations have shown that these patterns translate into reproductive success. In Polish (Pawlowski et al., 2000), North American (Mueller & Mazur, 2001), and British (Nettle, 2002a) populations, tall men have more children because they are less likely to remain childless and more likely to have second families. Thus, the patterns of sexual behavior we observe in modern human societies may indeed influence reproductive success, although modern practices of contraception undoubtedly constrain any response to selection.

Attractive traits can certainly be altered by grooming practices and need not be entirely honest signals of mate quality. However, some of the preferences found here could well be adaptations for finding high-quality mates. For example, masculine male faces honestly advertise health during adolescence (Rhodes et al., 2003) and average male faces advertise health during childhood (Rhodes et al., 2001), and both traits were associated with male mating success. Male preference for feminine traits may also be an adaptation for finding high-quality mates, although the case here is weaker because the preference was restricted to faces, whereas the association with health and fertility is clearer for bodies (Singh, 1993) than for faces (Rhodes et al., 2003). Overall, our results suggest that attractiveness plays an important role in mate choice for both men and women, and are consistent with the view that attractive traits and our preferences for them are under sexual selection (Fink & Penton-Voak, 2002, Rhodes & Zebrowitz, 2002, Thornhill & Gangestad, 1999).