Subclasses of receptors exist for most neurotransmitters. Frequently, two subtypes of receptors coexist in the same tissue and, in some cases, they mediate the same physiological response. In tissues with two classes of binding sites for a given hormone, an estimate of the proportion of each class of binding sites is obtained by inhibiting the binding of a single concentration of a radioligand with a selective unlabeled ligand. Accurate estimates of the density of each class of receptors will only be obtained, however, if the radioligand is entirely nonselective. Selectivity of just 2- to 3-fold can markedly influence the results of subtype analysis. The conclusion that a radioligand is nonselective is usually based on the results of a saturation binding curve. If Scatchard analysis of such data results in a linear plot, then it is concluded that the radioligand is nonselective. However, Scatchard analysis cannot distinguish between a radioligand that is nonselective and one that is slightly selective. The use of a slightly selective radioligand can lead to errors of 50% or more, depending on the concentration of the radioligand relative to the Kd values of the two classes of sites. A new analytical method has been developed that can be used to quantitate 2- to 3-fold differences in the affinity of two distinct classes of binding sites for a radioligand. This new approach requires that a series of inhibition experiments with a selective unlabeled ligand be performed in the presence of increasing concentrations of the radioligand. Analysis of the resulting inhibition curves, utilizing the mathematical modeling program MLAB on the PROPHET system, yields accurate estimates of the density of each class of receptor as well as the affinity of each receptor for the labeled and unlabeled ligands. This approach was used to determine whether 125I-iodopindolol shows selectivity for beta 1- or beta 2-adrenergic receptors. A series of inhibition curves was generated with the unlabeled ligands ICI 89,406 (beta 1-selective) and ICI 118,551 (beta 2-selective), using membranes prepared from C6 glioma cells. These cells contain both beta 1- and beta 2-adrenergic receptors. 125I-Iodopindolol was determined to be 3-fold selective for beta 2-adrenergic receptors. Since the sensitivity of this approach is superior to that of Scatchard analysis, it is likely that other radioligands, previously thought to be nonselective, will be shown to be selective when analyzed by this method.