If mania and depression are part of the same pathological processes, one would predict that episodes of one prospectively increase the odds of episodes of the other. The aim of the present study was to test this hypothesis. For comparison purposes, their relationship was contrasted to the relationship between mania and periods of psychosis. Exploratory analyses also tested the degree to which episodes of each occur with greater frequency over time (i.e., kindling). Participants for the present study came from the Suffolk County Mental Health Project (N = 628), a study of first-admission patients with psychosis. Of these participants, 144 met diagnostic criteria for bipolar I disorder and were analyzed for the current study. Results indicated that mania in a given month predicted depression the following month, even after controlling for other symptoms. The reverse, however, was not the case. Mania and psychosis, in contrast, were found to be robust predictors of one another from month to month. Effects were not due to treatment or demographic differences. These findings provide evidence that mania and depression are weakly related. In contrast, mania and psychosis are more closely linked. Findings are consistent with suggestions that psychiatric nosology regroup mania more closely with …