Fixed-ratio discrimination (FRD) training session-accuracy curves were constructed using first-order, nonlinear regression and probit analyses to determine maximal (asymptotic) accuracy and the number of sessions required to reach half-maximal accuracy. Increased FRD difficulty (reductions in the differences between the 2 fixed-ratio values to be discriminated) and a training parameter change each increased the number of sessions required to reach half-maximal accuracy and decreased maximal FRD accuracy (i.e., session-accuracy curves were shifted down and to the right) regardless of analysis procedure. These findings indicate that the above manipulations induced mixed competitive-noncompetitive inhibition of the rate of FRD learning. Microencephalic rats were more sensitive to increases in FRD difficulty, whereas control rats were more sensitive to the training parameter change.