Few individuals with arthritis are sufficiently active. We surveyed a convenience sample of exercisers (N = 134) to examine the utility of social cognitive theory variables, namely, self-regulatory efficacy, negative outcome expectations, and pain acceptance for predicting planned physical activity according to Weinstein’s two prediction suggestions. Logistic regression revealed, after controlling for pain intensity, self-regulatory efficacy, negative outcome expectations, and pain acceptance distinguished groups achieving/not achieving planned physical activity, p < 0.001 (28% variance). A second model adding past physical activity also predicted the groups, p < 0.001 (57% variance). This is one of the first arthritis studies examining planned physical activity relative to Weinstein’s recommendations.