The aim of the present study was to verify and extend a recent, isolated observation showing that, in rats, a moderate dose of morphine may induce either an increase or a decrease in preference for saccharin, the direction of the response depending apparently on the concentration of the sweetener. Two experiments were performed successively. First, we showed that the preference threshold for saccharin (0.3 mM, two-bottle procedure) of rats placed on a schedule of restricted water access was significantly decreased following injection of 1 mg/kg of morphine. In the second experiment, three groups of naive rats were submitted to the preference test but the concentration of saccharin solution was different for each group, namely 0.3, 1 and 1.7 mM. After stabilization of the baseline responses the effect of morphine (1 mg/kg) was tested in each of the 3 groups. As observed previously morphine decreased the preference of the rats tested with the 0.3 mM solution, but markedly increased the preference of the two other groups tested with the 1 and 1.7 mM solutions respectively. The effects of low doses of naloxone (0.01, 0.1 and 1 mg/kg) were then tested on the same groups of rats with the same saccharin concentrations. The 0.01 mg/kg dose of the antagonist increased the preference for the groups of rats tested with the 0.3 and 1 mM solutions. The other two doses of naloxone decreased saccharin intake whatever the saccharin concentration used. It is suggested that these apparently paradoxical effects of morphine and naloxone could result either from the stimulation of opioid autoreceptors or from the differential stimulation of different opioid receptor subtypes.