Selection for fast growth rate was practiced at 3 different temperatures in inter- and intra-strain crosses of Neurospora. The selection procedure was specifically designed to test genetic models in relation to genetic progress. Estimates of the parameters involved in the theoretical response curves were obtained from the experimental data. An estimate of the genetic component of variance based only the early generations of selection was chosen over one which weighted each generation equally. Since environmental variances increased with temperature, data over temperatures were not pooled. Logarithmic and square root transformations of the data failed to stabilize the environmental variances among the temperatures or among the crosses within any one temperature. An investigation of the estimators involved in the predicted response curve indicated that estimates of the slope and intercept were least squares estimators conditional on the fact that genetic and environmental estimates were the true values. Relatively good agreement was obtained between observed and theoretical responses to selection at the 3 different temperatures.