The present experiments further characterized intravenous self-administration of the short-acting benzodiazepine midazolam in rats under conditions of unlimited access to the drug. The results of the first experiment demonstrated that rats responded at higher rates on a lever that produced an infusion of 0.05mg midazolam than on a lever that did not result in reinforcement, and that they transferred responding to the other lever when the reinforcement contingencies were reversed. These results provide further evidence of the reinforcing effects of midazolam. In the second experiment, rats responding at stable rates for 0.05mg midazolam per infusion exhibited increased responding when transferred to a lower dose (0.0125mg/infusion) and decreased responding when transferred to a higher dose (0.20mg/infusion) of the drug. However, the inverse relationship between responding and drug dose was apparent only during the first transfer session, after which no consistent relationship was observed. In support of previous observations, all rats exhibited a temporal pattern of responding for midazolam over the 24h sessions, with maximal responding occurring during the dark phase of the 12h light/dark cycle. However, the present results also provide preliminary evidence that rats given prolonged access to midazolam (more than 49 days) develop a more constant within session pattern of responding for midazolam. This may be related to the development of physical dependence and reflect an attempt to avoid withdrawal effects by maintaining a stable intake of midazolam within a session.