Background: Meta-analyses of randomised trials generated estimates of cardiovascular risk with non-steroidal anti-inflammatory drugs (NSAIDs) in high doses and sometimes atypical settings. Objectives: To update estimates of cardiovascular risks associated with NSAIDs prescribed in general populations. Methods: Systematic review and meta-analysis of community-based controlled observational studies published 2000–2010. We pooled adjusted relative risk (RR) estimates for cardiovascular events with individual NSAIDs, in different doses, in populations with low and high cardiovascular risk. NSAIDs were compared in pair-wise analyses generating relative relative risks (RRR). Results: Eligible studies: 28 case-control (184,946 cardiovascular events); 23 cohort (>2.3 million individuals). Of extensively-studied drugs, RR was increased at low doses with rofecoxib (less than diclofenac 1.3 (1.1, 1.5), celecoxib 1.4 (1.2, 1.6), RR day, >1200 mg/day, RR 1.8 (1.4, 2.3). Naproxen was risk-neutral. Etoricoxib, RR 2.1 (1.5, 2.9), etodolac, 1.6 (1.3, 1.9), indomethacin 1.3 (1.2, 1.4), and meloxicam 1.2 (1.1, 1.3) had elevated risks. In pair-wise comparisons, etoricoxib had significantly higher risk than ibuprofen (RRR 1.68; 99%CI 1.1, 2.5) and naproxen (1.75; 1.2, 2.6). Naproxen had lower RRR than ibuprofen (0.92; 0.87, 0.99). Relative risks were constant irrespective of background cardiovascular risk. Conclusions: Etodolac, indomethacin and meloxicam have received little scrutiny in randomized studies. The pharmacoepidemiological data suggest they increase cardiovascular risk. Etoricoxib concerns are reinforced. Extensive data on diclofenac augment anxiety about its safety; it is available without prescription in several countries but is associated with increased risk even at low doses. Neither high nor low doses of naproxen appear to increase risk giving it an apparent advantage over other NSAIDs.