The incidence of postnatal depression is high, and dramatic changes in steroid hormones and prolactin occur in the postpartum period. In an attempt to correlate these events, 147 mothers, six to eight weeks after delivery of a healthy infant, completed standard psychological tests, including the Edinburgh, Montgomery-åsberg, and Raskin scales. They also provided matched samples of plasma for assay of cortisol, oestradiol, progesterone and prolactin, and saliva for assay of cortisol and progesterone. All steroid concentrations were within the appropriate normal ranges. Of the mothers, 14.9% were depressed on all three scales. Significant correlations were seen between depression ratings and salivary progesterone and prolactin. In bottle-feeders, salivary progesterone was positively associated with depression, whereas in breast-feeders it was negatively associated. Plasma prolactin levels were inappropriately low in depressed breast-feeders. These data indicate that differing therapies may be appropriate for depression in breast- and bottle-feeders.