Six ruminally cannulated male sheep (Santa Ines × Dorper; 35 ± 0.9 kg BW) were used to evaluate the effects of diets containing different levels of starch and neutral detergent-soluble fiber on protozoa and bacterial fractions, as well as methane and carbonic gases productions. Lambs were assigned to a replicated 3 × 3 Latin square design. Each experimental period consisted of 20 d adaptation and 1 d sampling. Diets consisted of 30% corn silage and 70% concentrate (corn grain, citrus pulp, sunflower meal, soybean hulls, sunflower oil, urea and minerals), providing one diet with high starch content (26.7%), one with same content of starch and neutral detergent-soluble fiber (18.4%) and one with high neutral detergent-soluble fiber (24.3%). Ruminal content were sampled on d 21of each experimental period, at 1.5; 6.5 and 11.5 h after the morning feeding. Samples were strained through a nylon filter to separate liquid and solid phases. Liquid-associated protozoa (LAP) and bacteria (LAB) were obtained using series of filtrations, dilution with salt solution and centrifugations. Particle-associated bacteria (PAB) were obtained using agitation with a salt solution, as well as filtrations and centrifugations. Data were analyzed using the MIXED procedure of SAS, with repeated measures. In vitro fermentation was performed using 250-mL flasks in a completely randomized design with 3 treatments and 6 replicates to evaluate the methane and carbonic gasses production. Samples of diets were added at 1.25 g DM/flask, and consisted of same diets fed to ruminal fluid donor. Data were analyzed using the GLM procedure of SAS. There were no effect (P > 0.05) of diets on LAP (980.4 mg/kg DM), LAB (858.9 mg/kg DM) or PAB (1237.7 mg/kg DM). The methane production was unaffected by diets (7.1 L/kg DM; P > 0.05) but the high-starch diet decreased carbonic gas production (P < 0.05). The starch or neutral detergent-soluble fiber diets do not affect ruminal microbial populations nor methane production.