The efficiency of rumen microbial production (EMP) in vitro and in vivo was examined for three roughages (lucerne (Medicago sativa L.) hay, oat (Avenia sativa L.)–berseem clover (Trifolium alexandrinum cultivar BigBee) hay and maize (Zea mays L.) crop residue (MCR)) and for five isonitrogenous (106 g crude protein (N × 6·25)/kg) diets formulated from lucerne hay, oat–berseem clover hay, MCR, soya-bean meal and maize grain to provide degradable intake protein for the production of 130 g microbial protein/kg total digestible nutrients. EMP in vivo was determined by intestinal purine recovery in sheep and ranged from 240 to 360 g microbial biomass/kg organic matter truly degraded in MCR and in one of the diets respectively (P<0·05). EMP in vitro was estimated by the substrate degraded: gas volume produced thereby (termed partitioning factor, PF (mg/ml)) at times of estimated peak microbial production and after 16·0 and 24·0h of incubation. For the diets, PF values were significantly related to EMP in vivo at peak microbial production (P = 0·04), but not after 16·0 (P = 0·08) and 24·0h (P = 0·66). For roughages, PF values were significantly related to EMP in vivo only when measured after 16·0 h (P = 0·04). For MCR and diets, a close non-linear relationship was found between PF values at peak microbial production and EMP in vivo (R2 0·99, P<0·0001) suggesting a maximum EMP in vivo of 0·39. Low gas production per unit substrate degraded (high PF) was associated with high EMP in vivo. The in vitro study of the products of fermentation, short-chain fatty acids, gases and microbial biomass (by purine analysis) after 16·0h of incubation showed very strong relationships (R2 ≥ 0·89, P<0·0001) between short-chain fatty acids, gases and gravimetrically measured apparent degradability. Except for maize grain, the true degradability of organic matter estimated by neutral-detergent solution treatment agreed with the sum of the products of fermentation (R2 0·81, P=0·0004). After 16·0h of incubation, the synergistic effects of diet ingredient on diets were greater for microbial biomass (18%) than for short-chain fatty acids and gas production (7 %). It is concluded that measurement of gas production only gives incomplete information about fodder quality; complementation of gas measurements by true degradability measurements is recommended.