We aimed to explore the potential link of serum vitamin D level with nonalcoholic fatty liver disease (NAFLD). PubMed, Embase and the Cochrane Library database were searched until the end of February 2018. Clinical studies with sufficient data investigating the relationship between serum vitamin D and NAFLD were included. The outcome data were processed to make an overall estimate of combined standardized mean differences (SMD) and pooled odds ratio (OR)/hazard ratios with 95% confidence intervals (CIs). Of the 309 initially retrieved studies, 15 studies of high quality involving a total of 20 096 participants (including 7803 NAFLD patients) were included in this meta-analysis. Meta-analysis of continuous data indicated that NAFLD patients had averagely 0.90 ng/ml lower levels of 25-hydroxyvitamin D compared with the non-NAFLD subjects (SMD -0.90; 95% CI: -1.29 to -0.52). Parallelly, pooled dichotomous data revealed that serum vitamin D level is negatively associated with NAFLD (OR = 0.64, 95% CI = 0.54-0.77), albeit with substantial heterogeneity. Next, subgroup analysis showed that Western NAFLD patients were more likely to be vitamin D deficient (OR = 0.60, 95% CI = 0.46-0.78). Finally, meta-regression showed that sample size, ethnic background, and diagnosis of NAFLD were possible sources of heterogeneity in the meta-analysis. Our results revealed that serum vitamin D level was inversely associated with an increased risk of NAFLD. Patients with hypovitaminosis D might benefit from extra supplement of vitamin D against the risk of NAFLD.