The multi-enzymatic method with constant pH described by Dufour-Etienne et al. (Rec. Méd. Vét. 168, 789-796, 1992) was tested using 17 industrial dry dog foods and two ingredients, sodium caseinate and beef meat, which were used as references. The quantity of sodium hydroxide added to the enzymatic solution was measured after incubation times of 2, 5 and 10 min and the best prediction of the apparent protein digestibility was obtained with an incubation time fixed at 2 min. The volume of sodium hydroxide 0.1 N constantly added to maintain the pH of the solution at 8 was correlated to the value of in vivo digestibility of the protein measured in the dog. The correlation coefficient obtained was 0.90 with a standard error of the regression of 1.9 for all samples. The correlation coefficients calculated for incubation times of 5 and 10 min were lower than that at 2 min; thus an incubation period of 2 min provided the highest correlation coefficient for this multi-enzymatic method at constant pH. However, considering the lack of samples with an apparent protein digestibility ranging between 85 and 95%, the slope of the regression line and the correlation coefficient were strongly influenced by the ingredients, such that when the correlation coefficient was calculated for 17 dog foods without the reference ingredients, a coefficient of 0.71 (+/- 1.9) was obtained. These tests of in vitro digestibility are valuable to predict the apparent digestibility of the protein in the dog, but, it is essential to confirm the results by an in vivo test.