In electrically driven guinea pig left atria, positive inotropic responses to (−)-isoprenaline and the selective β 1-adrenoceptor agonist RO363 were obtained in the absence and in the presence of the functional antagonists adenosine, carbachol, gallopamil, nifedipine, and Ro 03-7894. Each of the functional antagonists reduced the maximum response to both agonists and produced nonparallel right-ward shifts in the cumulative concentration effect curves. For both agonists, dissociation constants (k a) were calculated using the equation described by Furchgott (1966) for irreversible antagonism. For RO363, which is a partial agonist with high agonist activity, the equations outlined for functional interaction by Mackay (1981) were also employed to calculate k a values. The k a values obtained by each method were compared with the dissociation constants (K D) for the two agonists determined from their ability to displace the radioligand (−)-[ 125l]iodocyano-pindolol from β 1-adrenoceptors in guinea pig left atrial membrane preparations. The estimates of k A varied substantially from K D values. The K D values were taken as more accurate estimates of the true values for the dissociation constants because a high degree of correlation exists between pK D and pD 2 values for a number of other β-adrenoceptor agonists that behave as partial agonists and between pK D and pK B values for a number of β-adrenoceptor antagonists. Thus it appears that there are serious limitations in the current theory for using functional antagonism as a means of obtaining agonist dissociation constants.