BACKGROUND: Antibiotic resistance is an urgent global problem, but reversibility is poorly understood. We examined the development and decay of bacterial resistance in community patients after antibiotic use. METHODS: This was a systematic review and meta-analysis. PubMed, EMBASE and CENTRAL (from inception to May 2017) were searched, with forward and backward citation searches of the identified studies. We contacted authors whose data were unclear, and of abstract-only reports, for further information. We considered controlled or times-series studies of patients in the community who were given antibiotics and where the subsequent prevalence of resistant bacteria was measured. Two authors extracted risk of bias and data. The meta-analysis used a fixed-effects model. RESULTS: Of 24,492 articles screened, five controlled and 20 time-series studies (total 16,353 children and 1461 adults) were eligible. Resistance in Streptococcus pneumoniae initially increased fourfold after penicillin-class antibiotic exposure [odds ratio (OR) 4.2, 95% confidence interval (CI) 3.5-5.4], but this fell after 1 month (OR 1.7, 95% CI 1.3-2.1). After cephalosporin-class antibiotics, resistance increased (OR 2.2, 95%CI 1.7-2.9); and fell to (OR 1.6, 95% CI 1.2-2.3) at 1 month. After macrolide-class antibiotics, resistance increased (OR 3.8, 95% CI 1.9-7.6) and persisted for 1 month (OR 5.2, 95% CI 2.6-10.3) and 3 months (OR 8.1, 95% CI 4.6-14.2, from controlled studies and OR 2.3, 95% CI 0.6-9.4, from time-series studies). Resistance in Haemophilus influenzae after penicillins was not significantly increased (OR 1.3, 95% CI 0.9-1.9) initially but was at 1 month (OR 3.4, 95% CI 1.5-7.6), falling after 3 months (OR 1.0, 95% CI 0.5-2.2). Data were sparse for cephalosporins and macrolides. Resistance in Enterobacter increased post-exposure (OR 3.2, 95% CI 0.9-10.8, from controlled studies and OR 7.1, 95% CI 4.2-12, from time-series studies], but was lower after 1 month (OR 1.8, 95% CI 0.9-3.6). CONCLUSIONS: Resistance generally increased soon after antibiotic use. For some antibiotic classes and bacteria, it partially diminished after 1 and 3 months, but longer-term data are lacking and urgently needed. TRIAL REGISTRATION: PROSPERO CRD42015025499 .