Direct observation(s) of energy intake (EI) via buffet meals served in the laboratory are often carried out within short-term exercise intervention studies. The reproducibility of values obtained has not been assessed either under resting control conditions or post-exercise, in overweight and obese females. A total of fourteen sedentary, pre-menopausal females (BMI 30·0 (sd 5·1) kg/m2) completed four trials; two exercise and two control. Each trial lasted 24 h spanning over 2 d; conducted from afternoon on day 1 and morning on day 2. An exercise session to expend 1·65 MJ was completed on day 1 of exercise trials, and three buffet meals were served during each trial. Reproducibility of post-exercise changes in energy and macronutrient intakes was assessed at each individual buffet meal by intraclass correlation coefficient (ri). Only the ri values for post-exercise changes in energy (ri 0·44 (95 % CI − 0·03, 0·77), P = 0·03) and fat intake (ri 0·51 (95 % CI 0·04, 0·81), P = 0·02) at the lunch buffet meal achieved statistical significance; however, these ri values were weak and had large associated 95 % CI, which indicates a large degree of variability associated with these measurements. Energy and macronutrient intakes at the breakfast and evening buffet meals were not reproducible. This study concludes that the frequently used laboratory-based buffet meal method of assessing EI does not produce reliable, reproducible post-exercise changes in EI in overweight and obese women.