The pharmacokinetic and pharmacodynamic interaction between azithromycin (CAS 83905-01-5), an azalide antibiotic, and midazolam (CAS 59467-70-8), a short-acting hypnotic agent, was investigated in an open, three-way cross-over study, including erythromycin (CAS 114-07-8) as a positive control. Twelve healthy male and female subjects had standard doses of azithromycin (500 mg o.d. over 3 days), or erythromycin (500 mg t.i.d. over 5 days), or no pretreatment. On the day of the last dose, they ingested 15 mg midazolam. Blood samples were collected and psychometric tests performed. Erythromycin pretreatment (E) significantly changed the pharmacokinetics of midazolam compared to control (C), whereas azithromycin (A) had no such effect. The parameters are summarized as follows: area under the concentration-time curve, AUC (C) 173.8 h.ng.ml-1 vs. (E) 662.7 h.ng.ml-1*+ and (A) 220.0 h.ng.ml-1; concentration maxima (C) 67.2 ng.ml-1 vs. (E) 182.3 ng.ml-1*+ and (A) 86.7 ng.ml-1; elimination half-life (C) 2.21 h vs. (E) 4.85 h* and (A) 2.41 h (* p < 0.05 vs. (C), +p < 0.05 vs. (A)). Pharmacodynamic tests (digit symbol substitution test; critical flicker fusion test; subjective analog scale for rating of alertness; duration of sleep) consistently showed significant differences after erythromycin pretreatment compared to control, but not after azithromycin. Erythromycin, but not azithromycin, causes clinically significant changes in the pharmacokinetics and pharmacodynamics of midazolam.