The ruminal digestibility of dietary ingredients is frequently estimated with degradation parameters (e.g., rate and extent of degradation). Mean degradation parameters (e.g., those in a feed library) are often used, but limited data suggest considerable variation around these means, potentially leading to imprecise digestibility estimates. This experiment quantified degradation parameter variation for forage hays and determined the impact of this variation on the precision of ruminal digestibility estimates. Degradation data were those previously published by our laboratory and included degradation rate, lambda(d) (h(-1)), fraction instantly degraded, a (g x g(-1)), potential extent of degradation, (a + b) (g x g(-1)), and fraction not instantly degraded that is potentially degradable, b (g x g(-1)) of alfalfa, grass, and grass-legume hays. Ruminal digestibilities of chemical fractions (dry matter, neutral detergent fiber, acid detergent fiber, hemicellulose, crude protein) were estimated using these data. Ninety-five percent confidence limits of digestibility were determined using propagation of uncertainty with measured standard deviations of degradation parameters. Values for coefficients of variation of degradation parameters were large; averaged across chemical fractions, they were 24.8, 28.6, 20.7, and 12.6% for lambda(d), a, b, and (a + b). Ninety-five percent confidence limits of digestibility were large (80.5% of digestibility means) and often overlapped each other, even when digestibility means differed greatly numerically. Consequently, digestibility values computed with mean degradation parameters may have little biological and practical significance. When uncertainty in all parameters but lambda(d) was set to zero (lambda(d) alone had uncertainty), 95% confidence limits still encompassed 54.5% of digestibility means. Thus, uncertainty in lambda(d) alone caused considerable imprecision in estimated digestibility. These results caution against using mean degradation parameters to estimate digestibility.