The levels of 3 alpha-hydroxy-5 alpha-pregnan-20-one (allopregnanolone) and the epimeric 3 alpha-hydroxy-5 beta-pregnan-20-one (pregnanolone) were studied in women with prospectively confirmed premenstrual syndrome (n = 15) and in a group of asymptomatic control women (n = 12) during the luteal phase of the menstrual cycle. Single late luteal phase plasma samples were selected to make comparisons of plasma hormone levels between patients and controls in the following measures: allopregnanolone, pregnanolone, the ratio of allopregnanolone to pregnanolone, the ratio for each of these anxiolytic steroids to the parent compound progesterone, and the ratio of the sum of allopregnanolone and pregnanolone to progesterone. Differences in these measures were compared by analysis of variance. Additionally, correlations were performed among the various hormone measures and between the hormone measures and the symptom self-ratings. Analysis of variance showed no significant between group differences in the plasma levels of allopregnanolone, pregnanolone, and progesterone. Plasma levels of both allopregnanolone and pregnanolone were correlated with plasma progesterone levels. However, there were no significant correlations between the severity of mood and behavioral symptoms and plasma levels of progesterone, allopregnanolone, and pregnanolone. These data suggest that symptoms of premenstrual syndrome are not associated with a simple deficiency state of either progesterone or its anxiolytic steroid metabolites.