To compare the safety and efficacy of outpatient and inpatient treatment of pyelonephritis in pregnancy. We performed a randomized controlled trial of pregnant women with pyelonephritis before 24 weeks' estimated gestational age, comparing inpatient and outpatient treatment. Sixty inpatients received cefazolin intravenously until afebrile for 48 hours, and 60 outpatients received two injections of ceftriaxone intramuscularly. All patients completed a 10-day course of oral cephalexin. We performed a urine culture 5–14 days after completion of therapy. The two groups were similar with respect to age, parity, temperature, estimated gestational age, initial white blood cell count, and incidence of bacteremia. Escherichia coli was the major uropathogen isolated (86% of cultures, 95 of 111). Twelve percent (13 of 111) of bacteria were resistant to cefazolin. Eleven outpatients and 12 inpatients had positive urine cultures after therapy (relative risk 0.9, 95% confidence interval 0.4–1.9). Three patients in each group had recurrent pyelonephritis. We switched six inpatients to gentamicin because of a worsening clinical picture (two) or a prolonged febrile course (four); no outpatients required a change in antibiotic (Fisher exact test, P = .03). One preterm delivery occurred in an inpatient with recurrent pyelonephritis. Outpatient antibiotic therapy is effective and safe in selected pregnant women with pyelonephritis.