Groups of naive rats were offered morphine sulfate for self-admininstration in doses of 0.0032-10 mg/kg for 6 days. On day 7 saline was substituted for morphine. Loss of weight was taken as physiological evidence of dependence. Rats that did not lose weight formed a single population whose mean injection rate did not differ from control rats receiving only saline injections. Injection rates for rats losing weight were log-normally distributed, and the mean of the logarithms of the injection rates was linearly related to the logarithm of the dose. Mean daily injection rates averaged 12 for controls, 23 at 10 mg/kg, and 411 at 0.01 mg/kg. A transient increase in morphine intake after an injection of nalorphine was taken as behavioral evidence of dependence. Nalorphine increased morphine intake when rats were self-injecting 0.32 and 1.0 mg/kg of morphine, but not 0.032 or 0.1 mg/kg. The reinforcing property of morphine may occur without behavioral evidence of dependence.