The technique of chronic schedule-induced drug solution intake was used to determine the possible addiction liability of the short-acting benzodiazepine midazolam. Schedule-induction produces polydipsia over a wide range of fluids as a function of the imposed schedule of food availability. The inducing schedule used presented food pellets automatically once per minute, fixed time (FT) 1-min, for 3 h daily. Two groups of rats, drinking either water or 0.05 mg/ml midazolam solution, were exposed to schedule-induction sessions for approx. 2 months. Then, other FT-values (0, 0.5, 3 and 5 min) were instituted on occasion for single sessions. Each of these 'probe' session determinations was done twice. Although midazolam concentration had been adjusted so that the mean group intakes were equal at FT 1-min, probe values greater than 1 min revealed a greater acceptance of midazolam compared to water. This technique produced session midazolam intakes as great as 25 mg/kg. In the next phase, the entire experiment was repeated except both groups were offered a choice between water and midazolam solution during sessions. Only at FT 0 and FT 5-min was there an indication that midazolam was preferred over water. Two additional groups of animals were exposed to the same schedule-induced polydipsia regimen, drinking water and midazolam solution, respectively. Pre-session administration of doses of Ro 15-1788, CGS 8216 (benzodiazepine antagonists) or midazolam had no effects on either water or midazolam intakes, although the higher dose of midazolam (2 mg/kg, s.c) had a moderately suppressive effect on the non-tolerant water-polydipsic group. All groups were tested on occasion for physical dependence on midazolam with an auditory stimulus as the precipitator and midazolam polydipsic groups were found to have a mild to moderate dependence (clonic seizures, running fits).