Recent work suggests that climate change will impact negatively on Canadian prairie wheat through more severe and frequent droughts and increased yield variability, and adaptive strategies have been called for to meet the coming change. However, previous studies have not determined if climate change has already affected wheat yields, nor has the entire prairie region been examined. Using a prairie wheat yield simulation model (PWYSM) at current (constant) technology, it is shown that the weather as it affects wheat, has had a zero trend over the 1900–1988 period. Also, there is no evidence that yield variability has recently increased. Prairie wheat yields therefore appear so far unconnected to the global warming trend this century, suggesting no current need to adapt to a changing climatic regime. Results also show that, at current technology, departures from mean weather may cause decade-long average yield fluctuations as large as ± 10% from the long-term mean, and climate change effects must be detected against this background fluctuation. Also, a 35% decline in actual wheat yields from the first to fourth decades of this century was not matched by deteriorating weather although weather contributed to the decline, and improved yields since 1940 are not explained by better weather. The study of Williams et al. (1988) concluded that a 1930s type dry spell would reduce Saskatchewan wheat yield by 20%, compared to the 1961–1979 mean That applied only to stubble-sown wheat. When adjusted here for the contribution of fallow-sown wheat, the reduction became 16.7%. With Alberta and Manitoba included, the yield reduction for the whole prairie became 13.8% — in good agreement with a 14.0% reduction obtained with the PWYSM. Yield reductions for Saskatchewan are larger than for the prairie as a whole. The reference period chosen also significantly affects findings, and, because the last decade has been relatively unfavourable, PWYSM results show that 1930s type weather would lower mean prairie wheat yield by only 11.0% relative to the 1979–1988 mean. Grain industry decision makers would more likely refer to the last decade than to a more distant period such as 1961–1979, and applying relative yield reductions to the 1979–1988 mean would result in substantial yield underestimates for drought scenarios. Reference periods must be carefully selected for consistent results. Key words: Climate change, wheat yield