A rate-frequency curve in self-stimulation experiments plots the response rate as a function of the frequency of stimulation pulses, yielding a steeply rising, roughly sigmoidal curve. Altering stimulation reward efficacy results in lateral shifts of this curve, while altering the operant performance capacity of the rat results in primarily vertical shifts. One hypothesis explaining the lateral stability of the curve in the face of performance-altering manipulations is that the stimulation reward effect saturates at the frequency where the curve asymptotes. In the first experiment we determined the frequency of rate-frequency curve asymptote in two paradigms, runway and lever-pressing, and compared it to the reward saturation frequency determined in a discrete choice procedure. Results indicate that reward often saturates at frequencies above the rate-frequency asymptote point, which does not support the above hypothesis; levelling-off of the rate-frequency function is most likely a result of performance ceiling factors. In a second experiment, increases in stimulation current reduced the saturation frequency, indicating that saturation is not determined by an upper limit on the signal carrying capacity of the directly excited axons.