Crossbred sheep (n 16, 8.5 months of age and 33 (SE 0.9) kg) were used in a 21 d experiment (2 x 2 factorial) to determine effects on net flux of nutrients across the portal-drained viscera (PDV) and liver of ad libitum consumption of bermudagrass (Cynodon dactylon; B) v. ryegrass (Lolium multiflorum)-wheat (Triticum aestivum; RW) hay, coarsely chopped (CC) or finely ground and pelleted (GP). Crude protein concentrations were 86, 81, 113 and 119 g/kg and neutral-detergent fibre concentrations were 710, 688, 654 and 672 g/kg (dry matter basis) for B-CC, B-GP, RW-CC and RW-GP respectively. Digestible energy intake (6.0, 9.6, 10.2 and 13.8 MJ/d) differed (P < 0.01) with grass source and form, and digestible N intake values were 4.4, 7.0, 8.4 and 14.1 (SEM 0.82) g/d for B-CC, B-GP, RW-CC and RW-GP diets respectively. Consumption of O2 by the PDV (118, 165, 144 and 155 mmol/h) and splanchnic bed (196, 273, 247 and 266 mmol/h for B-CC, B-GP, RW-CC and RW-GP respectively) was greater (P = 0.07) for GP than for CC. The ratio splanchnic heat energy production: digestible energy intake was greater (P = 0.06) for B than for RW (0.374, 0.300, 0.278 and 0.219 for B-CC, B-GP, RW-CC and RW-GP respectively). alpha-Amino-N release by the PDV (P < 0.01; 11.6, 12.8, 23.0 and 18.7 mmol/h) and uptake by the liver (P = 0.07; 15.2, 6.1, 17.0 and 19.3 mmol/h for B-CC, B-GP, RW-CC and RW-GP respectively) were greater for RW than for B. Release of NH3-N by the PDV was greater (P = 0.02) for CC than for GP (12.5, 6.2, 15.7 and 8.9 mmol/h), and hepatic urea-N release differed between grass sources (P = 0.03) and physical forms (P = 0.07; 22.6, 12.7, 31.4 and 24.8 mmol/h for B-CC, B-GP, RW-CC and RW-GP respectively). In conclusion, decrease in forage particle size elicited by grinding and pelleting did not affect the difference between grass sources in splanchnic tissue heat energy production relative to digestible energy intake.