In vivo and in situ protein degradation measurements were compared using alfalfa (2.62% N) and prairie hay (.88% N) fed to six cannulated cows in a two-period crossover experiment. Additionally, two in situ procedures were evaluated: in P1 samples were ruminally incubated in cows fed the same forage as incubated; in P2 samples were incubated in the rumens of two steers fed a "standard" brome hay (1.38% N). Duplicate bags were incubated for 0, 2, 6, 10, 16, 24, 48, and 72 h. Protein degradability was estimated using residual N from all incubation times fit to a nonlinear, least squares model (full time-series), from the 16-h incubation alone (single-point) and from a combination of the 0-h and 16-h values (double-point). Protein degradability estimates from in vivo, in situ P1, in situ P2, single-point P1 and P2, and double-point P1 and P2 were 83.4 +/- 4.3, 91.5 +/- .6, 87.2 +/- .6, 94.0 +/- .2, 92.4 +/- .8, 90.4 +/- .4, and 88.3 +/- .5, respectively, for alfalfa, and 55.5 +/- 3.5, 58.3 +/- 1.3, 57.2 +/- .4, 50.0 +/- 1.7, 52.0 +/- 2.2, 55.5 +/- 2.1, and 60.0 +/- 2.3, respectively, for prairie hay. Although relatively large differences in rates of degradation and sizes of protein fractions were measured between P1 and P2, differences between procedures for protein degradability estimates were small relative to errors of in vivo measurement. Furthermore, differences in protein degradation using single-point values as compared with full time-series analysis were overcome with the double-point approach.