1 Recent theoretical studies have questioned the pharmacological estimation of agonist affinity. They showed that when receptor isomerisation or ternary complex mechanisms operate, the receptor inactivation method can substantially overestimate affinity, whereas methods for partial agonist analysis are more accurate. We previously suggested that the operation of such mechanisms and therefore the presence of errors could be detected by analysing the same partial agonist by the receptor inactivation and comparative methods. This paper describes the practical application of this test. 2 The ternary complex mechanism was simulated for a partial agonist under various conditions relating receptor (R) and transducer (T) concentrations, one of which also corresponds to the receptor isomerisation mechanism. The theoretical data so generated were then analysed by the inactivation-and comparative methods to quantify the magnitude of error of affinity estimation that could occur. 3 This analysis showed that for a partial agonist with approximately 85% of the activity of a full agonist, the inactivation method could produce an affinity (pKA) estimate up to 0.7 log1o units higher than that produced by the comparative method. This difference would occur when the total receptor concentration ([RO]) is less than or equal to the total transducer concentration ([To]). It also showed that the overestimation of affinity by the inactivation method was accompanied by drastic overestimation of Em, the maximal effect parameter. 4 The test was then exemplified using the muscarinic receptor system in the guinea-pig isolated left atrial preparation, where there is evidence that a ternary complex mechanism operates. The test agonist was pilocarpine, which produced on average 83% of the activity of the full agonist, carbachol. Pilocarpine was analysed in comparison with carbachol and by receptor inactivation in the same tissue resulting in small and statistically insignificant differences in Em (96.7% and 97.3% respectively) and pKA (5.03 and 4.95 respectively). 5 In conclusion, in this experimental system, there was no evidence for the errors in agonist affinity estimation predicted by theory. Although this conclusion only applies to this system and application of the test to others is necessary to establish the generality of the present results, further examination of the theoretical basis for the predicted errors is required.