The parents chosen to continue 10 independent selection lines of Schizophyllum commune over eight successive generations of selection, along with unselected controls, have been retrospectively examined for their response to growth at 15 degrees, 20 degrees, 25 degrees, 30 degrees and 35 degrees C. The regression of rate of growth on temperature was essentially linear over the range 15 degrees to 30 degrees C for all lines in all generations as was also the regression of rate of growth on various biological assessments of the environments over the whole temperature range. Either regression, therefore, provided linear regression coefficients which adequately accounted for the relative sensitivities of the lines to temperature in each generation of selection. These measures of environmental sensitivity confirmed our earlier report that selection for high mean performance in a good environment or for low mean performance in a poor environment leads to selections that are more sensitive to environmental variation than selections for high mean performance in a poor environment or for low mean performance in a good environment. These differences in sensitivity emerge as correlated responses during selection and the magnitude of these correlated responses is higher in the good environment than in the poor environment irrespective of the direction of selection. The environmental sensitivity of selection lines can be modified in either direction as required by either selecting for sensitivity simultaneously with the selection for mean performance or by selecting for mean performance in an above or below average environment. The quality of environments in which artificial selection is usually carried out is likely to have led to high selections with maximum environmental sensitivity and low selections with minimum sensitivity.