The responses of wheat to various rates of N fertiliser were compared following field peas (PW) or wheat (WW) in the previous year. Seventeen trials were carried out at 5 sites between 1986 and 1991. The trials were on medium- and fine-textured soils (clay loams or shallow duplex soils). The overall grain yield of PW appeared greater than WW in 11 trials [was significantly greater in 9 (P<0.05)], and did not appear different in 6 trials. When no N was applied the yield advantage of PW was 41% (PW 1.91 t/ha cf. WW 1.37 t/ha). Quadratic response curves were fitted to all yield data. Rotation x N rate interaction was significant (P<0.05) in 10 comparisons. In 5 trials, while there was a yield increase to N fertiliser with WW, the yields decreased with PW. In 3 trials while there was an increase with WW there was no response with PW or a reduction at higher rates of N. In the remaining 2 trials there were responses with both PW and WW, but this was greater for WW. The response curves in these 10 trials either converged and met, indicating that the difference between rotations was due to N availability, or converged but did not meet, indicating that N was important but did not explain the whole difference. Where there was no interaction between rotation and N rate the response curves were parallel. The type of response could not be predicted. It was not profitable to apply N fertiliser to wheat in the PW rotation in 11 of the 17 trials. The average yield advantage of PW over WW, in the absence of N was 540 kg/ha, while there was an average increase of 1.7% grain protein.