The aim of this systematic review and meta-analysis was to assess the effect of exercise in population-based interventions to prevent low back pain (LBP) and associated disability. Comprehensive literature searches were conducted in multiple databases, including PubMed, Embase, and the Cochrane Library, from their inception through June 2017. Thirteen randomized controlled trials (RCTs) and 3 nonrandomized controlled trials (NRCTs) qualified for the meta-analysis. Exercise alone reduced the risk of LBP by 33% (risk ratio = 0.67, 95% confidence interval: 0.53, 0.85; I2 = 23%, 8 RCTs, n = 1,634), and exercise combined with education reduced it by 27% (risk ratio = 0.73, 95% confidence interval: 0.59, 0.91; I2 = 6%, 6 trials, n = 1,381). The severity of LBP and disability from LBP were also lower in exercise groups than in control groups. Moreover, results were not changed by excluding the NRCTs or adjusting for publication bias. Few trials assessed health-care consultation or sick leave for LBP, and meta-analyses did not show statistically significant protective effects of exercise on those outcomes. Exercise reduces the risk of LBP and associated disability, and a combination of strengthening with either stretching or aerobic exercises performed 2-3 times per week can reasonably be recommended for prevention of LBP in the general population.