Arachidonylethanolamide (anandamide), an endogenous ligand for the cannabinoid receptor, binds competitively to brain cannabinoid receptors and shares many, but not all, of the in vivo effects of delta9-tetrahydrocannabinol. In this study, the cannabinoid effects of anandamide analogs in which the anandamide molecule was altered were assessed in a drug discrimination model. Structural manipulations of the anandamide molecule included saturation of the arachidonyl moiety with fluorination (O-586), substitution for either the ethanolamide moiety (O-612 and O-595) or C2' hydroxyl (O-585), and addition of a methyl group at various positions (O-610, O-680, and O-689). Despite the low binding affinities of the non-methylated compounds (Ki values > 2000 nM), all of the analogs had previously shown cannabinoid activity in mice. In the present study, these analogs were tested in a more pharmacologically specific delta9-tetrahydrocannabinol discrimination procedure in rats. This animal model is predictive of the subjective effects of marijuana intoxication in humans. Whereas delta9-tetrahydrocannabinol and an aminoakylindole fully substituted for the training dose of 3 mg/kg delta9-tetrahydrocannabinol, anandamide and its non-methylated analogs were not cannabimimetic in this procedure. Methylation appeared to increase binding affinity (Ki values < 150 nM) and efficacy; however, the greatest substitution produced by the methylated analogs occurred only at doses that decreased overall rates of responding, suggesting that these analogs are not fully delta9-tetrahydrocannabinol-like. The rapid metabolism of anandamide and some of its analogs undoubtedly contribute to the differences between the pharmacological profiles of the anandamides and classical cannabinoids. These results support the prediction that the subjective effects of anandamide analogs that have been developed thus far would not be cannabimimetic except at high doses.