We tested the hypothesis that low plasma adiponectin is associated observationally and causally with increased risk of type 2 diabetes. Observational analyses are prone to confounding and reverse causation, while genetic Mendelian randomization (MR) analyses are much less influenced by these biases. We examined 30,045 individuals from the Copenhagen General Population Study observationally (plasma adiponectin [1,751 individuals with type 2 diabetes]), 96,903 Copenhagen individuals using one-sample MR (5 genetic variants [5,012 individuals with type 2 diabetes]), and 659,316 Europeans (ADIPOGen, GERA, DIAGRAM, UK Biobank) using two-sample MR (10 genetic variants [62,892 individuals type 2 diabetes]). Observationally, and in comparisons with individuals with median plasma adiponectin of 28.9 μg/mL (4th quartile), multivariable adjusted hazard ratios (HRs) for type 2 diabetes were 1.42 (95% CI 1.18-1.72) for 19.2 μg/mL (3rd quartile), 2.21 (1.84-2.66) for 13.9 μg/mL (2nd quartile), and 4.05 (3.38-4.86) for 9.2 μg/mL (1st quartile). Corresponding cumulative incidence for type 2 diabetes at age 70 years was 3%, 7%, 11%, and 20%, respectively. A 1 μg/mL lower plasma adiponectin conferred an HR for type 2 diabetes of 1.07 (1.06-1.09), while genetic, causal risk ratio per 1 unit log-transformed lower plasma adiponectin was 1.13 (95% CI 0.83-1.53) in one-sample MR and 1.26 (1.01-1.57) in two-sample MR. In conclusion, low plasma adiponectin is associated with increased risk of type 2 diabetes, an association that could represent a causal relationship.