The suitability of the pig as an animal model for predicting protein digestibility in man was evaluated. Healthy adult human subjects (mean body weight 67 kg; n 11) and growing pigs (mean body weight 40 kg; n 15) were fed semi-synthetic mixed meals containing, as a sole source of N, casein (C), hydrolysed casein (HC) or rapeseed isolate (R). There was no prior adaptation to the test meal. Ileal digesta were sampled through a naso-ileal tube (human subjects) or a post-valve T-caecum cannula (pigs) after ingestion of a bolus meal. The protein sources were 15N-labelled. Amino acid (AA) digestibilities were not determined for R. Ileal apparent N digestibility was markedly lower (14–16 %; P < 0·001) in human subjects than in pigs (C, HC, R). Similarly, most apparent ileal AA digestibilities were lower (8 % on average; P < 0·05) in human subjects (C, HC). Ileal true N digestibility was slightly lower (3–5 %; P < 0·001) in human subjects than in pigs (C, HC, R) and most true ileal AA digestibilities were similar (P>0·05) between the species (C, HC). Exceptions were for phenylalanine, tyrosine, lysine, histidine and aspartic acid for which digestibilities were lower (3 % on average; P < 0·001) in human subjects. A similar ranking of the diets was observed for true ileal N digestibility between species. The inter-species correlation for true ileal digestibility was high for N (r 0·98 over 3 × 2 data; P = 0·11) and AA (r 0·87 over 26 × 2 data; P < 0·0001). Overall, this supports the use of the pig as a model for predicting differences among dietary protein digestibility, especially regarding true ileal N digestibility, in man.