Objective To evaluate the efficacy and safety of rifampicin plus pyrazinamide versus isoniazid for prevention of tuberculosis among persons with or without HIV-infection respectively. Methord Meta-analysis of randomized controlled trials(RCT) and quasi-randomized controlled trials(quasi RCT) that compared rifampicin plus pyrazinamide for 2-3 months with isoniazid for 6-12 months. Endpoints were development of active tuberculosis, severe adverse effects, and death. Treatment effects were summarized as risk difference (RD) with 95% confidence interval (CI). Results Three trials conducted in HIV-infected patients and 3 trials conducted in HIV-uninfected persons were identified. The rates of tuberculosis in the rifampicin plus pyrazinamide group were similar to that in the isoniazid group, whether the subjects were HIV-infected patients or not (for HIV-infected patients: pooled RD= 0%, 95%CI: -1% to 2%, P=0.89; for HIV-uninfected persons: pooled RD=0%, 95%CI: -2% to 1%, P=0.55). There was no difference in mortality between the two treatment groups (for HIV-infected patients: pooled RD=-1%, 95%CI: -4% to 2%, P=0.53; for HIV-uninfected persons: pooled RD=0%, 95%CI: -1% to 1%, P=1.00). However, both subgroup analyses showed that a higher incidence of all severe adverse events was associated with rifampicin plus pyrazinamide than isoniazid among HIV-uninfected persons (one: RD=29%, 95%CI: 13% to 46%; P=0.000 5; another: RD=7%, 95%CI: 4% to 10%; Plt;0.000 1). Conclusion Rifampicin plus pyrazinamide is equivalent to isoniazid in terms of efficacy and mortality in the treatment of latent tuberculosis infection. However, this regimen increases risk of severe adverse effects compared with isoniazid in HIV-uninfected persons.