The dopamine D1-like receptor agonists have traditionally been defined molecularly by their efficacy in stimulating adenylyl cyclase. However, evidence correlating the effectiveness of these drugs in behavioral assays and their effectiveness biochemically has not been forthcoming. The present study compared the discriminative-stimulus effects of the D1-like partial agonist SKF 38393 with several other D1-like agonists, an indirect agonist, cocaine, and a D2-like agonist, quinpirole. Rats were trained under a fixed-ratio 30-response schedule to discriminate SKF 38393 (5.6 mg/kg) from vehicle. Under this schedule, 30 consecutive responses on one of two keys were reinforced with food presentation after a pre-session injection of 5.6 mg/kg SKF 38393, and 30 consecutive responses on the alternative key were reinforced after saline injection. When daily performances were stable, substitution patterns for several compounds were assessed during test sessions in which 30 consecutive responses on either key were reinforced. Quinpirole and cocaine each produced saline-appropriate responding. In contrast, the D1-like agonists, SKF 75670 and SKF 77434, fully substituted for SKF 38393. Curiously, SKF 82958, which is considered a full agonist based on adenylyl cyclase assays, was less effective in substituting for SKF 38393 (maximum drug-appropriate responding 66%) than was the partial agonist SKF 75670. The present results suggest that second messenger effects other than stimulation of adenylyl cyclase may play an important role in the behavioral effects of dopamine D1-like agonists.