Abstract Objectives To evaluate the risk of serious infection (SI) and herpes zoster (HZ) in rheumatoid arthritis patients receiving JAK inhibitors. Methods We conducted a systematic literature review and meta-analysis of phase II and III randomized controlled trials of tofacitinib (5 mg bid), baricitinib (4 mg od) and upadacitinib (15 mg od). Patient-exposure years were calculated. A per-protocol analysis was applied, incorporating follow-up time from patients randomized to placebo who cross into the treatment arm. Pooled incidence rates per 100 person-years of SI and HZ were calculated. Incidence rate ratios (IRRs) of drug vs placebo were compared using a meta-synthesis approach. Results Twenty-one studies were included in the meta-analysis; 11 tofacitinib (5888 patients), six baricitinib (3520 patients) and four upadacitinib studies (1736 patients). For SI, the incidence rates were 1.97 (95% CI: 1.41, 2.68), 3.16 (95% CI: 2.07, 4.63) and 3.02 (95% CI: 0.98, 7.04), respectively. The IRRs comparing treatment arm to placebo were statistically non-significant: 1.22 (95% CI: 0.60, 2.45), 0.80 (95% CI: 0.46, 1.38) and 1.14 (95% CI: 0.24, 5.43), respectively. For HZ, the incidence rates were 2.51 (95% CI: 1.87, 3.30), 3.16 (95% CI: 2.07, 4.63) and 2.41 (95% CI: 0.66, 6.18), respectively. The IRR of HZ comparing baricitinib with placebo was 2.86 (95% CI: 1.26, 6.50). Non-significant IRRs were seen with tofacitinib and upadacitinib: 1.38 (95% CI: 0.66, 2.88) and 0.78 (95% CI: 0.19, 3.22), respectively. Indicator opportunistic infections excluding HZ were too rare to provide meaningful incidence rates. Conclusion The absolute SI rates were low. However across the JAK inhibitors, the incidence of HZ is higher than expected for the population (3.23 per 100 patient-years). While the risk was numerically greatest with baricitinib, indirect comparisons between the drugs did not demonstrate any significant difference in risk. Systematic review registration number Prospero 2017 CRD4201707879.