The comparative tolerability and rate of withdrawal from clinical trials was established from a pooled analysis of comparative clinical studies of roxithromycin and/or erythromycin in patients with lower respiratory tract infections. Where stated, 65% of patients were managed in the community and some as inpatients. 942 patients in 13 studies received roxithromycin 300 mg/day and 1253 patients in 15 studies received erythromycin in various formulations and dosages. 47% of patients had an initial diagnosis of community-acquired pneumonia, 20% had an acute infectious exacerbation of chronic bronchitis, 30% had acute bronchitis, and a further 4% had unspecified lower respiratory tract infection. The incidence of adverse events that were definitely or probably associated with antibacterial therapy (excluding laboratory abnormalities) was 2.5-fold higher for patients receiving erythromycin compared with roxithromycin (24.8 vs 10.1%), and the rate of withdrawal from clinical trials due to adverse events was 3.6-fold higher (7.1 vs 2.0%). Most of the adverse events associated with erythromycin were gastrointestinal in nature. Since adverse events and withdrawals help determine the compliance rate and the clinical value of an antibacterial agent in routine clinical practice, they should be taken into account when prescribing a macrolide for a respiratory tract infection.