To test a causal model derived from discrete affect (Tomkins, 1962) and involvement (Mosher, 1980) theories, 120 male and 121 female subjects responded to four erotic conditions with self‐reports of discrete affects and subjective sexual arousal. The four erotic conditions were (a) recall of a past heterosexual episode, (b) recall of a past masturbatory episode, (c) free sex fantasy, and (d) guided sex fantasy. The model posits (a) a direct effect of sex on sex guilt only; (b) sex guilt and masturbation guilt would have direct effects and indirect effects on sexual arousal; and (c) discrete positive affects would amplify sexual arousal, discrete negative affects would attenuate sexual arousal, and discrete affects would mediate the indirect effects of sex guilt on sexual arousal. Covariance structure analysis (Duncan, 1975; Kenny, 1979) was used to generate path coefficients. Incremental fit indices indicated an excellent goodness of fit of the model. Sex had no significant paths. Sex guilt had both direct and indirect effects on sexual arousal. Discriminant validity was a problem in measuring negative affects. Improvements in methods of measurement and experimental design are suggested for future research.