This study has modelled an alternative approach to systems of supply response by decomposing it into acreage and yield responses for Western Canada grains, using aggregate data across provinces for 1961-1995. This study follows a similar study for Manitoba by Mundang (1996). Acreage demands are estimated by employing duality models and allowing risk aversion. Econometric results are presented for a four-crop model using level and first differenced data. Estimates for models incorporating risk aversion generally have more significant coefficients for expected revenues or expected prices (under price uncertainty). Results for models assuming yield uncertainty and nonlinear estimations are less satisfactory than in earlier studies. This may be due to difficulties in consistent aggregation across provinces and that the model does not capture changes in agricultural policies that may influence acreage demand decisions. The yield responses to price are specified assuming price uncertainty, risk aversion, and distributed lags. Under static models, the expected price coefficients are always insignificant, however, price variances are often significant (wheat). In contrast, under distributed lag models, both expected prices and price variances are often significant. This results suggest that it is important to incorporate risk aversion and dynamics into crop yield response models. For further research, it would be interesting to specify a model for each province and to incorporate changes in agricultural policies more explicitly into the model.