This study investigated the relationships of pain and sleep with depressive symptoms in women with rheumatoid arthritis. Specifically, the study aimed to replicate the findings of Nicassio and Wallston (1992), and extend their investigations by examining the hypothesised relationships over multiple time intervals. A further aim was to explore the role of social activity in predicting depressive symptoms. Subjects were 81 women with a diagnosis of definite rheumatoid arthritis. Each woman completed questionnaires 12 times, at 3‐to 4‐month intervals. Data from the first five time‐points are used in the current analyses. At Time 1, pain was significantly associated with depressive symptoms, and there was also a trend for low social activity to be associated with concurrent depressive symptoms. Longitudinally, the interaction of either sleep with social activity or pain with social activity at Time 1 predicted depression 12 months later. When multiple time intervals were used in the analysis, disability was the only significant predictor of depressive symptoms. The implications of these results are disc