We investigated the predictability of infradian mood fluctuations during acute depressive episodes in patients affected by mood disorders. Previous findings showed that in a subgroup of patients the depressive symptomathology fluctuates with day-to-day changes which follow cyclical patterns (termed "minicycles'). We applied time series analysis, by means of autocorrelation techniques, to time lagged serial recordings of perceived mood levels of 22 inpatients (13 Major Depressive Recurrent and 9 Bipolar Depressive Disorders). Five patients (22.7%) were shown to exhibit predictable cyclical patterns in their perceived symptomathology, ranging in length from 6 to 14 days. Our study confirms the existence and the predictability, in a subgroup of patients, of cyclical mood patterns. Preliminary evidence suggests that patients with minicycles receive more medication changes than patients without, and thus that cyclical mood fluctuations strongly interacts with both the clinical decision making process and the outcome of acute depressive episodes.