Starch digestibility is often estimated in vitro at 7 h and the value is then used to calculate a rate of digestion using a previously published mathematical formula. The objective of this study was to compare mathematical approaches for determining the rate of starch digestion using data from various grains and particle sizes. Three cereals (corn, sorghum, barley) were ground through a 2-mm sieve using a Wiley mill and subsequently sieved to obtain the following sizes: <250, 250–500, 500–1180 and 1180–2000 µm. All fractions, including the unsieved one, were analyzed for starch and fermented in vitro using rumen fluid to quantify starch digestibility (Sd). Residual starch of the fermented samples were obtained at 0, 3, 6, 7, 9, 12 and 24 h. Rates of starch digestion (kd) were calculated with either a first-order decay model using all time points except 7-h (nlin-kd), or a mathematical formula using only the 7-h Sd (7h-kd). The nlin-kd’s were used as reference values (observed) and compared with the 7h-kd values. Digestibility values, pooled, by grain or by size, were pairwise-compared and prediction accuracy was tested and compared using correlations and the mean square prediction error (MSPE). Fermentation run (n = 3) was considered as random effect. Predicted and observed kd were highly correlated across all grains and sizes (r = 0.94 to 0.99). Across grains, corn resulted in the highest correlation (r = 0.98) while sorghum and barley resulted in r = 0.77 and 0.87, respectively. While observed kd regressed on predicted kd resulted in intercepts not different than zero (P > 0.44), slopes were always larger or smaller than 1 (P < 0.01) resulting in the 7-h kd under- or over-estimating, respectively, the reference nlin-kd. The formula resulted in underestimated values especially for the unsieved samples, with differences up to 0.04 h⁻¹. This suggests the need of using a nonlinear estimation, using multiple time points, or the development of alternative estimations, especially when quantifying rates of starch digestion for high producing cows.