This study evaluated the effect of pooling rumen fluid (RF) or using fluid of one cow across incubations on in vitro gas production (GP) on parameters of feedstuffs. A complete randomized block design was used and arranged in a 4 × 2 factorial scheme. Concentrate (corn grain, barley, soybean meal, mineral and vitamin premix) grass silage, grass hay and grass pasture were dried, ground and placed into 160-mL glass bottles. Duplicates of each substrate (1g) plus 2 blanks were incubated for three 48-h runs (considered as technical replicates). Bottles with 85 mL of Goering-Van Soest buffer solution were purged with CO₂ and 10 mL of RF was added to them. One treatment was incubated with RF from one cow (different for each run; not-pooled), whereas the other, a pool of RF from the same 3 cows was used for all runs. The volume of gas was extracted with a syringe, until the pressure in the digital display of the transducer reached 0, at times 2, 3, 4, 5, 6, 8, 10, 12, 18, 24 and 48h after incubation. A generalized Michaelis-Menten (MM) model was used to estimate fermentation kinetics, considering A: asymptote of GP (mL/g OM); N: shape of the curve; C: half-life rate degradation (%/h); K: time to ferment 50% of the substrate (h); MDR: maximum degradation rate (%/h); t25, t75: time to ferment 25 and 75% of asymptotic GP, respectively (h); 48GP: GP at 48 h. MIXED procedure of SAS was used, with the incubation method and substrate as fixed, and the run as random effect. The variance components (substrate, incubation, and error) were determined for each method with the varcomp procedure of SAS. There was no interaction between substrates and inoculation method; therefore, the variables were analyzed independently. The use of pooled RF promoted greater values of A (218) and t75 (23.3) compared with not-pooled RF (A = 212, t75 = 21.4), with P values and SED of 0.02 and 2.3, and 0.03 and 0.8 for A and t75, respectively. All other parameters did not differ between inoculation methods. Substrate effect was significant (P < 0.001) for all MM parameters. For both inoculation methods, more than 94% of the variance was due to the substrates for all MM parameters. The variance accounted for incubation-runs was similar among inoculation method. Gas production parameters of different substrates follow a same trend regardless of the inoculation method.