Ninety-four patients with bipolar disorder participating in a random-assignment, double-blind, prospective maintenance trial of standard- (0.8 to 1.0 mmol/L) vs low-range (0.4 to 0.6 mmol/L) serum lithium levels were assessed to determine the presence and significance of subsyndromal symptoms during periods of remission and recovery. A significant relationship was found between prescribed serum lithium level and the probability of major affective relapse and the occurrence of subsyndromal symptoms. Patients given lithium carbonate to achieve low-range levels had 2.6 times the risk of major affective relapse as those given lithium for standard-range levels and nearly twice the risk of developing subsyndromal symptoms. Patients given the low-range therapy showed a greater variance in weekly Psychiatric Status Rating measures, and their symptoms were more likely to worsen at any time than were symptoms in their standard-level group counterparts. The first occurrence of subsyndromal symptoms increased the risk of major affective relapse fourfold. Following the onset of subsyndromal symptoms, the patients originally randomized to receive standard-range lithium therapy were still better protected from relapse than were patients randomized to receive low-range lithium treatment. Patients were two times more likely to develop depressive than hypomanic symptoms between acute episodes of illness. However, onset of hypomanic symptoms predicted subsequent major affective relapse twice as strongly as did depressive symptoms. Seventy-six percent of patients who became hypomanic had a major affective relapse, compared with 39% of patients who were subclinically depressed.