A series of esters of 6beta-hydroxynortropane and the N-methyl analogue 6beta-tropanol were synthesized and screened versus binding of an antagonist (quinuclidinyl benzilate) and an agonist (oxotremorine-M) at sites on human m(1)-, m(2)-, m(3)-, and m(4)-muscarinic receptors in transfected cell membranes and on native M(1)-muscarinic receptors in rat brain membranes and native M(2)-muscarinic receptors in rat heart membranes. Most 6beta-acyloxy(nor)tropanes had higher affinity versus oxotremorine-M binding compared to quinuclidinyl benzilate binding at transfected m(1)- and native M(1)-receptors, indicative of agonist activity. 6beta-Acetoxynortropane had K(i) values versus oxotremorine-M binding at m(1)-, m(2)-, and m(4)-receptors ranging from 4 to 7 nM. N-Methylation reduced affinity greatly as did increasing the size of the acyl moiety. The affinity of 6beta-benzoyloxynortropane and other analogues with larger acyl moieties was little affected by N-methylation or in some cases was increased. 6beta-Acyloxy(nor)tropanes and classical muscarinic agonists, such as muscarine and oxotremorine, had higher affinity versus oxotremorine-M binding compared to quinuclidinyl benzilate binding at native M(2)-muscarinic receptors of heart, but not at transfected m(2)-muscarinic receptors. Antagonist/agonist binding ratios were not obtained for transfected m(3)-receptors, since significant oxotremorine-M binding could not be detected. 6beta-Acyloxy(nor)tropane, two other (nor)tropanes, and the classical muscarinic agonists had higher affinity versus agonist binding compared to antagonist binding for transfected m(4)-receptors. The antagonist/agonist binding ratio method is clearly not always reliable for predicting agonist activity at muscarinic receptors.