Background There are distinct results for the relationship between new-onset atrial fibrillation (NOAF) and subsequent incident cancer. To date, no systematic analysis has been conducted on this issue. This study aims to explore the relationship between NOAF and the risk of developing cancer through a meta-analysis with a large sample size. Methods Electronic databases, such as PubMed and EMBASE, were searched for published relevant studies on NOAF patients diagnosed with cancer after and during follow-ups, including reported records of baseline information and the statistical result of morbidity. Two investigators independently reviewed the articles and extracted the data using uniform standards and definitions. The meta-analysis was conducted using the Cochrane Program Review Manager. Results This meta-analysis consisted of five cohort studies and one case-control study, which comprised of 533,514 participants. The pooled relative risk (RR) for incident cancer was 1.24 (95% CI: 1.10-1.39, P=0.0003). The temporal trends analysis demonstrated that an increased risk of cancer was observed during the initial 90 days (RR: 3.44, 95% CI: 2.29-5.57, P<0.00001), but not after that. Lung cancer (RR: 1.51, 95% CI: 1.47-1.55, P<0.00001) was associated with NOAF, but not colorectal cancer and breast cancer. Conclusion This meta-analysis provides evidence that NOAF is associated with increased risk of cancer. The risk of incident cancer particularly increases within 90 days after NOAF diagnosis, but not after that.