A trial was carried out in general practice in 200 patients presenting with skin infections to compare topical antibiotic treatment with mupirocin ointment with orally administered flucloxacillin or erythromycin. Patients were assigned at random to receive 4 to 10 days' treatment with either mupirocin applied 3-times daily or one of the oral antibiotics in the dosage normally used by the general practitioner for skin infections. The majority of infections were impetigo and infected wounds/lacerations; the main organisms isolated initially from 127 of the patients were either Staphylococcus aureus or beta-haemolytic Group A streptococci. Clinical response to mupirocin ointment (86% cured, 13% improved) was significantly better than that seen with erythromycin (47% cured, 26% improved) and similar to that with flucloxacillin (76% cured, 23% improved). Treatment outcome was not related to treatment duration with either the topical or oral preparations. Post-treatment samples from 76 patients showed that in the mupirocin group all the pathogens originally isolated were eliminated, including Gram-negative organisms.