Urea recycling to the gut allows ruminants to maintain a positive N balance as recycled urea can support ruminal microbial growth. Our objective was to determine the effects of forage level and site of starch digestion on N utilization and in vitro serosal-to-mucosal urea flux (J_sm-urea) across ruminal, duodenal and cecal epithelia using sheep as a model. Wether lambs (32) were blocked by BW into groups of 4 and randomly assigned (within blocks) to 1 of 4 isoproteic (10.7% CP) diets (n = 8) in a 2 × 2 factorial design. Diet factors were forage level (30% [LF] vs. 70% [HF]) and corn grain processing (whole-shelled [WS] vs. steam-flaked [SF]). Feeding SF was expected to shift starch digestion from post-ruminal sites to the rumen when compared with WS. Four blocks of lambs (n = 4) were used to determine N balance. Lambs were killed after 23 d of diet adaptation. Ruminal, duodenal and cecal epithelia were collected and mounted in Ussing chambers to determine J_sm-urea. Forage level, grain processing and their nteractions were tested using Proc Mixed of SAS. Lambs fed SF had greater (P ≤ 0.03) DM (1.23 vs. 1.04 kg/d) and N (20.6 vs. 14.5 g/d) intakes than those fed WS. Lambs fed LF tended to have a greater DMI (1.21 vs. 1.06 kg/d; P = 0.07) and had a greater N intake (NI; 20.1 vs. 15.0 g/d; P < 0.01) than those fed HF. When expressed as g/d, total N excretion was greater (P < 0.01), whereas when expressed as a % of NI, it was lower (P = 0.02) in lambs fed SF compared with WS. Retained N was greater in lambs fed SF compared with WS (P = 0.02). Total N excretion (g/d or % of NI) was greater (P ≤ 0.02), whereas retained N was lower (P < 0.01), in lambs fed HF compared with LF. Ruminal pH was similar in lambs fed SF and WS with HF diets, but it was lower in lambs fed SF compared with WS with LF diets (interaction, P = 0.01). Lambs fed LF had a lower cecum pH compared with those fed HF (P < 0.01). Ruminal and cecal J_sm-urea fluxes were unaffected by diet. Duodenal J_sm-urea flux was greater in lambs fed HF compared with LF (77.5 vs. 57.2 nmol·(cm²)⁻¹·h⁻¹; P < 0.01). These results suggest that shifting starch digestion to the rumen by feeding SF compared with WS improves N retention.