Musical composers, instrumentalists, and painters were compared with nonmusicians from a student and from an nonstudent population on testosterone levels in saliva. This steroid served as a marker for physiological androgyny. The ANOVA showed a significant group by sex interaction. Male composers attained significantly lower mean testosterone values than male instrumentalists and male nonmusicians; female composers had significantly higher mean testosterone values than female instrumentalists and female nonmusicians. Painters of both sexes did not differ significantly from controls. Spatial ability was assessed in the five groups. Significant differences on spatial test performance were not reflected in differences on salivary testosterone. Our results showed that musical composers of both sexes were physiologically highly androgynous. Creative musical behavior was associated with testosterone levels that minimized sex differences.