Experimental probing of the model of the muscarinic M2 receptor binding site proposed by Hibert et al. [Hibert, M.F., Trumpp-Kallmeyer, S., Bruinsvels, A., Hoflak, K., 1991. Three-dimensional models of neurotransmitter G-binding protein-coupled receptors. Mol. Pharmacol. 40, 8-15.] was achieved by mutating each amino-acid proposed to interact with muscarinic ligands. Pharmacological analysis of the different mutant receptors transiently expressed in human embryonic kidney (HEK/293) cells was performed with a variety of agonists and antagonists. D103A, Y403A and N404A mutations prevented binding of [3H] N-methylscopolamine and [3H] quinuclidinyl benzilate with a reduction in affinity greater than 100-fold, indicating essential contributions of these residues to the binding site for the radioligands. W400A and W155A mutations had very large effects on the binding of [3H] N-methylscopolamine (150-fold, 960-fold) but modest effects on the binding of [3H] quinuclidinyl benzilate (4-fold, 17-fold). In addition, binding of oxotremorine-M, oxotremorine, arecoline and pilocarpine to W155A resulted in a greater than 100-fold decrease in affinity. Threonine mutations (T187A and T190A) alter binding of most agonists but not of antagonists. W99 makes little contribution (< 10-fold) to the binding site of the M2 receptor. D103, W155, W400, Y403 and N404 are likely to be part of the binding site for N-methylscopolamine and also to contribute to the binding site for quinuclidinyl benzilate. Some of the predicted residues do not seem to be part of the M2 receptor binding site but W155 is important for proper ligand binding on the muscarinic M2 receptor, as predicted by the proposed model.