OBJECTIVES: Cannabis is the most commonly used illicit substance in the United States. Usage rates have climbed in recent years, underscoring the need for knowledge regarding the effects of cannabis use on factors associated with chronic health problems, such as heart disease and diabetes mellitus. Some studies suggest that cannabis use is associated with improvements in weight, BMI, and insulin resistance.  METHODS: Data on 4,267 persons from Continuous National Health and Nutrition Examination Survey (NHANES) from 2005 to 2010 was used to explore the relationship between cannabis use and factors of metabolic syndrome, including fasting insulin, glucose, insulin resistance, hemoglobin A1c, triglycerides, HDL cholesterol, BMI, waist circumference, and blood pressure. These relationships were first estimated with ordinary least squares (OLS) models. Next, instrumental variables (IV) methods were utilized to test and account for the potential endogeneity of cannabis use in the models. The first IV models used sexual behavior variables as instruments for past and current use of cannabis. The second used past cannabis use as an instrument for current use.  RESULTS: OLS models show lower fasting insulin, insulin resistance, BMI, and waist circumference in past cannabis users compared to individuals who reported never having used cannabis. In the first IV model, the coefficients on cannabis use are mostly non-significant. When past cannabis use is an instrument for current use, the results for fasting insulin, insulin resistance, and fasting glucose are significant in the opposite direction from the OLS results. Durbin-Watson-Hausman tests provide evidence of endogeneity of cannabis use for some outcomes.  CONCLUSIONS: Models of the relationship between cannabis and health should account for endogeneity. Results of two-stage least squares estimation are inconsistent with OLS results, challenging the robustness of findings that indicate a positive relationship between cannabis use and fasting insulin, insulin resistance, BMI, and waist circumference.