Background: Neutrophil gelatinase-associated lipocalin (NGAL) is a promising marker for the diagnosis of diabetic kidney disease (DKD), but its utility is currently debated. This meta-analysis aims to evaluate the diagnostic value of NGAL for DKD. Method: MEDLINE, Embase, Cochrane Library, CNKI, and CBM databases were searched up to April 13, 2019. In bivariate random-effect models, the diagnostic performance of NGAL for DKD was assessed using pooled estimates of sensitivity, specificity, likelihood ratio, diagnostic odds ratio, and hierarchical summary receiver-operating characteristic analysis. Results: Nineteen studies were eligible for the meta-analysis. Serum NGAL had a pooled sensitivity and specificity of 0.79 (95% confidence intervals [CI] 0.60–0.91) and 0.87 (0.75–0.93) (7 studies, 1,238 patients). The pooled positive likelihood ratio (LR+) and negative likelihood ratio (LR–) were 5.97 (3.03–11.76) and 0.24 (0.11–0.51). For urine NGAL, the pooled sensitivity, specificity, LR+, and LR– were 0.85 (0.74–0.91), 0.74 (0.57–0.86), 3.26 (1.87–5.67), and 0.21 (0.12–0.35), respectively (10 studies, 1,369 patients). The pooled sensitivity and specificity for kidney disease in normoalbuminuric patients with diabetes was 0.90 (0.82–0.95) and 0.97 (0.90–0.99) for both serum NGAL and 0.94 (0.87–0.98) and 0.90 (0.81–0.96) for urine NGAL (4 studies, 221 patients). NGAL appeared to perform similarly in subgroup analysis. Conclusion: The meta-analysis has shown that NGAL may be useful for DKD classification and also has a potential diagnostic value for normoalbuminuric kidney disease. Large-scale prospective studies are required to clarify its role in the diagnosis and risk stratification of patients with DKD.