Most life stress literature in bipolar disorder (BD) fails to account for the possibility of a changing relationship between psychosocial context and episode initiation across the course of the disorder. According to Post’s (1992) influential kindling hypothesis, major life stress is required to trigger initial onsets and recurrences of affective episodes, but successive episodes become progressively less tied to stressors and may eventually occur autonomously. Subsequent research on kindling has largely focused on unipolar depression (UD), and the model has been tested in imprecise and inconsistent ways. The aim of the present paper is to evaluate evidence for the kindling model as it applies to BD. We first outline the origins of the hypothesis, the evidence for the model in UD, and the issues needing further clarification. Next, we review the extant literature on the changing relationship between life stress and bipolar illness over time, and find that evidence from the methodologically strongest studies is inconsistent with the kindling hypothesis. We then integrate this existing body of research with two emerging biopsychosocial models of BD: the Behavioral Approach System dysregulation model, and the circadian and social rhythm theory. Finally, we present therapeutic implications and suggestions for future research.