Macrolide antibiotics are known to inhibit the metabolism of triazolam and midazolam in vitro and in vivo. To find out if significant interactions take place after single oral doses of these agents to man, 0.25 mg triazolam and 5, 10 and 15 mg of midazolam in capsule from were given with and without 750 mg erythromycin or 300 mg roxithromycin to parallel groups of healthy subjects in four placebo-controlled double-blind studies. Objective tests and subjective assessments were made before the intake of hypnotics and 30 and 90 min after it. In Study I, triazolam impaired letter cancellation, the combination triazolam+erythromycin impaired digit symbol substitution and letter cancellation, and triazolam+roxithromycin impaired digit symbol substitution, all at 90 min. In Study II, midazolam 5 mg and midazolam 10 mg proved quite inert but the combination midazolam 5 mg+erythromycin impaired digit symbol substitution. In Study III, both midazolam 10 mg and midazolam 15 mg impaired digit substitution and letter cancellation, the effects of 15 mg being more prominent. The strongest drug effects were found with midazolam 10 mg+erythromycin which differed from placebo and midazolam (10 mg and 15 mg) in several objective and subjective test variables. In Study IV, the combination midazolam 10 mg+roxithromycin impaired several objective and subjective variables but it was not stronger than midazolam 15 mg. These results were supported by the direct measurements of plasma midazolam in three subjects: erythromycin increased plasma midazolam more than roxithromycin and enhanced midazolam effects following the intake of midazolam 10 mg.(ABSTRACT TRUNCATED AT 250 WORDS)