Seventeen experiments were conducted in 1996, 1997 and 1998 in the central and northern wheatbelt of Western Australia, covering a range of soil types, seasonal rainfall, cultivars and sowing times. The objective of the experiments was to investigate how these factors affect the range of optimum seed rates derived from seeding rate experiments and, thus, to improve advice to farmers. Our results suggest that soil type and seasonal rainfall were the major factors influencing the differences in optimum seed rate. Regression tree methods were used to show that experiments in clay loam soils had higher optimum seed rates (52-76 kg/ha, depending on the cultivars used). In sandier soils, the optimum seed rate was lower (35-60 kg/ha, depending on cultivar and sowing time) but higher (67 kg/ha) at higher seasonal rainfall (>450 mm). We found some cultivars were grouped into consistent response patterns. Sowing time also influenced optimum seed rate; later sowing required higher seed rates, to maximise grain yield. A positive correlation was not observed between grain yield and optimum seed rate, possibly due to the narrow range of yields recorded in the experiments. Our data showed that the percentage of establishment fell off rapidly at higher seed rates. This implies that lower establishment percentages should be used when calculating the seed rates required to produce high plant populations in the field.