The novel radioligand [3H]AF-DX 384 binds specifically and saturably to putative muscarinic M2 receptor sites in homogenates of rat cerebral cortex. In saturation studies, [3H]AF-DX 384 appears to bind to two subpopulations of sites/states, one of high affinity (Kd1 = 0.28 +/- 0.08 nM) and another of low affinity (Kd2 = 28.0 +/- 5.0 nM). The maximal binding capacity (Bmax) of [3H]AF-DX 384 binding sites represented 9.7 +/- 2.3 fmol/mg protein (Bmax1) and 1993 +/- 551 fmol/mg protein (Bmax2) for the high and low affinity sites/states, respectively. The ligand selectivity profile of [3H]AF-DX 384 (at 2 nM) revealed that (-)-quinuclidinyl benzylate = atropine greater than 4-diphenylacetoxy-N-methylpiperidine methobromide greater than AQ-RA 741 greater than AF-DX 384 greater than UH-AH 371 much greater than methoctramine greater than oxotremorine-M greater than hexahydro-sila-defenidol much greater than pirenzepine greater than carbamylcholine much much greater than nicotine. This suggests that under our assay conditions [3H]AF-DX 384 binds mostly to M2-like muscarinic receptors in the rat central nervous system. This is further supported by the clear M2-like pattern of distribution observed using quantitative receptor autoradiography. High densities of specific labelling were seen in areas such as the hypoglossal nucleus, the pontine nucleus, the superior colliculus, the motor trigeminal nucleus, various thalamic nuclei and certain cortical laminae. Compared to [3H]AF-DX 116, the percentage of specific binding detected with [3H]AF-DX 384 was much higher. This is likely to be related to the greater chemical stability and affinity of [3H]AF-EX 384. In addition, autoradiograms obtained with [3H]AF-DX 384 (2 nM) are of better quality with film exposure periods five shorter than those needed for [3H]AF-DX 116 (10 nM). Therefore, [3H]AF-DX 384 displays a good selectivity for muscarinic M2 sites and offers major advantages, including higher affinity and greater stability, over previously used ligands.