Sufficient protein intake has been suggested to be important for preventing physical frailty, but studies show conflicting results which may be explained because not all studies address protein source and intake of other macronutrients and total energy. Therefore, we studied 2504 subjects with data on diet and physical frailty, participating in a large population-based prospective cohort among subjects aged 45+ years (the Rotterdam Study). Dietary intake was assessed with a FFQ. Frailty was defined according to the frailty phenotype as the presence of at least three out of the following five symptoms: weight loss, low physical activity, weakness, slowness and fatigue. We used multinomial logistic regression models to evaluate the independent association between protein intake and frailty using two methods: nutrient residual models and energy decomposition models. With every increase in 10 g total, plant or animal protein per d, the odds to be frail were 1·06 (95 % CI 0·98, 1·15), 0·87 (95 % CI 0·71, 1·07) and 1·07 (95 % CI 0·99, 1·15), respectively, using the nutrient residual method. Using the energy partition model, we observed that the odds to be frail were lower with higher vegetable protein intake (OR per 418·4 kJ (100 kcal): 0·61, 95 % CI 0·39, 0·97), however, results disappeared when adjusting for physical activity. For energy intake from any source we observed that with every 418·4 kJ (100 kcal) increase, the odds to be frail were 5 % lower (OR: 0·95, 95 % CI 0·93, 0·97). Our results suggest that energy intake, but not protein specifically, is associated with less frailty. Considering other macronutrients, physical activity and diet quality seems to be essential for future studies on protein and frailty.