The effects of some macrolides (4 mmoles . kg-1 p.o. daily for 4 days in vivo; 0.3 mM in vitro) on hepatic drug-metabolizing enzymes in rats were compared. One group of macrolides including previously studied compounds (oleandomycin, erythromycin and troleandomycin), as well as several other erythromycin derivatives, showed induction of microsomal enzymes and formation of inactive cytochrome P-450-metabolite complexes in vivo; this formation increased in the order: oleandomycin, erythromycin ethylsuccinate, erythromycin stearate, erythromycin itself, erythromycin propionate, erythromycin estolate and troleandomycin. Troleandomycin and, to a lesser extent, erythromycin and oleandomycin formed cytochrome P-450-metabolite complexes when incubated in vitro with 1 mM NADPH and microsomes from rats pretreated with troleandomycin or phenobarbital, but not with microsomes from control rats or rats treated with 3-methylcholanthrene. In contrast, two other macrolides, josamycin and midecamycin, showed no induction of microsomal enzymes and no detectable formation of cytochrome P-450-metabolite complexes in vivo. In vitro, these macrolides failed to form detectable complexes even with microsomes from rats pretreated with troleandomycin or phenobarbital. Hexobarbital sleeping time was unaffected by preadministration of josamycin or midecamycin (4 mmoles . kg-1 p.o.) 2 hr earlier; the in vitro activity of hexobarbital hydroxylase was not inhibited by 0.3 mM josamycin or midecamycin. We conclude that, unlike several erythromycin derivatives, josamycin and midecamycin do not form inactive cytochrome P-450-metabolite complexes in rats.