Two plausible theories of agonist action - the isomerization and ternary complex mechanisms - predict that agonist affinity measured using the inactivation method may be subject to overestimation. Moreover, the greater the intrinsic efficacy, the greater the predicted error. But are these predictions correct, and are they borne out by experimental data? Paul Leff and colleagues argue that accurate agonist affinity constants can be measured despite these predictions, but that effort must be made to detect any potential errors that might be anticipated from theory.