Abstract Background Globally, most deaths due to childhood pneumonia occur at the community level. Some countries are still using oral co-trimoxazole, despite a World Health Organization recommendation of oral amoxicillin for the treatment of fast-breathing pneumonia in children at the community level. Methods We conducted an unblinded, cluster-randomized, controlled-equivalency trial in Haripur District, Pakistan. Children 2–59 months of age with fast-breathing pneumonia were treated with oral amoxicillin suspension (50 mg/kg/day) for 3 days in 14 intervention clusters and oral co-trimoxazole suspension (8 mg trimethoprim/kg and 40 mg sulfamethoxazole/kg/day) for 5 days in 14 control clusters by lady health workers (LHW). The primary outcome was treatment failure by day 4 for intervention clusters and by day 6 for control clusters. The analysis was per protocol. Results Out of the 15 749 cases enrolled in the study, 9153 cases in intervention and 6509 cases in control clusters were included in the analysis. Treatment failure rates were 3.6% (326) in intervention clusters and 9.1% (592) in control clusters. After adjusting for clustering, the risk of treatment failure was lower in intervention clusters (risk difference [RD] -5.5%, 95% confidence interval [CI] -7.4–-3.7%) than in control clusters. Children with incomplete adherence had a small increase in treatment failure versus those with complete adherence (RD 2.9%, 95% CI 1.6–4.1%). No deaths or serious adverse events occurred. Conclusions A 3-day course of oral amoxicillin, administered by LHWs, is an effective and safe treatment for fast-breathing pneumonia in children 2–59 months of age. A shorter course of amoxicillin improves adherence to therapy, is low in cost, and puts less pressure on antimicrobial resistance. Clinical Trials Registration ISRCTN10618300.