•Blood and urinary cadmium is linearly associated with type 2 diabetes risk.•Diabetes risk starts to increase above 1 µg/L of blood cadmium levels.•Prediabetes risk increases up to 2 µg/g of creatinine, then reaches a plateau. Cadmium exposure has been associated with increased diabetes risk in several studies, though there is still considerable debate about the magnitude and shape of the association. To perform a systematic review and meta-analysis of observational studies investigating the relation between cadmium exposure and risk of type 2 diabetes and prediabetes, and to summarize data on the magnitude and shape of the association. After conducting an online literature search through October 1, 2021, we identified 42 eligible studies investigating the association between cadmium exposure and risk of diabetes and prediabetes. We included studies that assessed cadmium exposure through biomarker levels; examined type 2 diabetes or prediabetes among outcomes; and reported effect estimates for cadmium exposure for meta-analysis only. Studies were evaluated using ROBINS-E risk of bias tool. We quantitively assessed the relation between exposure and study outcomes using one-stage dose-response meta-analysis with a random effects meta-analytical model. In the meta-analysis, comparing highest-versus-lowest cadmium exposure levels, summary relative risks (RRs) for type 2 diabetes were 1.24 (95% confidence interval 0.96–1.59), 1.21 (1.00–1.45), and 1.47 (1.01–2.13) for blood, urinary, and toenail matrices, respectively. Similarly, there was an increased risk of prediabetes for cadmium concentrations in both urine (RR = 1.41, 95% CI: 1.15–1.73) and blood (RR = 1.38, 95% CI: 1.16–1.63). In the dose-response meta-analysis, we observed a consistent linear positive association between cadmium exposure and diabetes risk, with RRs of 1.25 (0.90–1.72) at 2.0 µg/g of creatinine. Conversely for blood cadmium, diabetes risk appeared to increase only above 1 µg/L. Prediabetes risk increased up to approximately 2 µg/g creatinine above which it reached a plateau with RR of 1.42 (1.12–1.76) at 2 µg/g creatinine. This analysis provides moderate-certainty evidence for a positive association between cadmium exposure (measured in multiple matrices) and risk of both diabetes and prediabetes.