Canola (Brassica napus L.) is widely grown throughout all rainfall zones in south-western Australia. Yields are low by world standards, and variable in low-rainfall (<350 mm annual rainfall) and medium-rainfall (350–450 mm) zones, so that minimising production costs is a major consideration for growers in these areas. One of the major input costs is nitrogen (N) fertiliser. Fifteen N rate × application time × canola plant-type experiments were conducted in the low- and medium-rainfall zones between 2012 and 2014. In most experiments, five rates of N were tested, of ranges 0–75, 0–100, or 0–150 kg N/ha. Nitrogen was applied at four different times (seeding, or 4, 8 or 12 weeks after sowing) or split between these timings. Each experiment compared triazine-tolerant (TT), open-pollinated (OP) canola with Roundup Ready (RR) hybrid canola, and one experiment included TT hybrid and RR OP canola types. On average, RR hybrid produced 250 kg/ha, or 23% more seed and 2.2% more oil than TT OP canola, and the average gross margin of RR hybrid was AU$65/ha more than TT OP. However, seed yield and gross margin differences between RR hybrid and TT OP canola were reduced when seed yields were <1400 kg/ha.

Canola growth (dry matter) and seed yield responded positively to N fertiliser in most experiments, with 90% of maximum seed yield achieved at an average of 46 kg N/ha (s.e. 6). However, 90% of maximum gross margin was achieved at a lower average N rate of 17 kg N/ha, due primarily to the relatively small yield increase compared with the reduction in concentration of oil in the seed with N applied. Because canola growers of south-western Australia are now paid an uncapped premium for canola grain with oil concentration >42%, decreases in oil percentage have a significant financial effect, and recommended rates of N should be lower than those calculated to optimise seed yield. In 80% of cases, the first 10 kg N/ha applied provided a return on investment in N >$1.50 for every $1 invested. The next 20 kg N/ha applied provided a return on investment of $1.25 for every $1 invested 80% of the time, and further increases would most likely break even. The timing of N application had a minor effect on yield, oil and financial returns, but delaying N application would allow farmers to reduce risk under poor conditions by reducing or eliminating further inputs. Overall, our work demonstrates that a conservative approach to N supply mindful of the combined impacts of N on yield and oil is necessary in south-western Australia and that split and delayed applications are a viable risk-management strategy.