Recently, Zernig and colleagues (1996) (J Pharmacol Toxicol Meth 35: 223-237) suggested that for the estimation of agonist affinity and efficacy, the method of simultaneously fitting of concentration-effect curves from control and irreversible antagonist-treated tissues to the operational model of agonism is superior to other analytical approaches. In the present study, we have evaluated the limitations of this simultaneous curve fitting method. Simulation studies showed that this method can be only employed with confidence when the upper asymptotes of the control curves display minimal variation between tissues, which makes its practical utility rather limited. The unreliability of the simultaneous fitting procedure was further underscored with the analysis of experimental data obtained from the interaction between noradrenaline and phenoxybenzamine in rat isolated aorta. The lack of robustness of the parameter estimates showed that under standard experimental conditions the outcomes of simultaneous model fitting are highly dependent on between-tissue variations of the upper asymptotes of the control curves and, therefore, may be unreliable. Therefore, whenever possible, a multiple curve design should be adopted, in which control and treated curves are obtained in one tissue and provide enough information for an independent estimation of affinity and efficacy that is free of intertissue differences.