OBJECTIVE: To assess the relationship between depressive relapse and change in thyroid function in an exploratory post hoc analysis from a controlled maintenance evaluation of bipolar I disorder. METHOD: Mean thyroid-stimulating hormone (TSH) and outcome data were pooled from two 18-month, double-blind, placebo-controlled, maintenance studies of lamotrigine and lithium monotherapy. A post hoc analysis of 109 subjects (n = 55 lamotrigine, n = 32 lithium, n = 22 placebo) with serum TSH values at screening and either week 52 (+/-14 days) or study drop-out was conducted. RESULTS: Lithium-treated subjects who required an intervention for a depressive episode had a significantly higher adjusted mean TSH level (4.4 microIU/ml) compared with lithium-treated subjects who did not require intervention for a depressive episode (2.4 microIU/ml). CONCLUSION: Lithium-related changes in thyroid function are clinically relevant and should be carefully monitored in the maintenance phase of bipolar disorder to maximize mood stability and minimize the risk of subsyndromal or syndromal depressive relapse.