Background: Many adults in behavioral weight loss (BWL) programs find long-term weight loss challenging, particularly after treatment ends. High levels of physical activity (PA) during follow-up may promote long-term weight loss, but little is known about the individual variability of this effect. This study tested the hypotheses that the strength of the relationship between PA and long-term weight loss depends on extent of emotional eating and appetitive response to PA. Methods: Participants (N = 271) received 18 mo of BWL. PA was measured with tri-axial accelerometers at 18, 24, and 36 mo. Weight was measured in clinic at 36 mo. Participants completed the Emotional Overeating Questionnaire (EOQ) and a single-item measure of perceived appetitive response to PA at baseline and 6 mo. Baseline weight was controlled for in analyses. Multiple imputation was used for missing data. Results: Weight loss at 36 mo was 11.1% in participants engaging in >150 min/wk of PA at all three time points (i.e., 18, 24, and 36 mo) ("High PA"), vs. 5.2% in participants engaging in >150 min/wk of PA at one or two of those time points ("Mixed PA"), vs. 3.1% in participants engaging in <150 min/wk of PA at each time point ("Low PA") (p < .001). PA interacted with EOQ to predict 36-month weight loss (p = 0.02). Weight loss at 36-mo was as follows: High PA, Low EOQ = 13.6%; High PA, High EOQ = 8.6%; Low PA, Low EOQ = 3.2%; Low PA, High EOQ = 2.7%. The relationship between PA and 36-mo weight loss also depended on perceived effect of PA on appetite at 6 mo (p < .05). Among those with High PA, 36-mo weight loss was as follows: No Effect on Appetite = 19.4%; Controls Appetite = 9.3%; Increases Appetite = 6.3%. Among those with Mixed PA, 36-mo weight loss was as follows: No Effect on Appetite = 6.1%; Controls Appetite = 5.4%; Increases Appetite = 3.1%. Conclusions: Specialized treatment tools may be warranted to maximize the benefit of PA for individuals who experience emotional eating or perceive that PA effects their appetite.