Obesity has been suggested as the third most important factor for cancer risk. Convincing evidence for an association with obesity, exists for at least oesophageal adenocarcinoma as well as colorectal-, pancreatic- postmenopausal breast-, endometrial-, kidney, and liver cancer. By estimating the population attributable fractions (PAFs) of all cancers it has been shown that at least 3.6 % of all cancers can be directly attributed to high body mass index (BMI), a fraction that is much larger in females compared to males. The global prevalence of obesity and overweight have risen dramatically over the last century and today over 124 million children and adolescents, and 670 million adults are obese. In an observational study, i.e. where the risk of cancer is compared between obese and non-obese individuals, the fact that an undetected cancer could influence obesity (reverse causation) cannot be disregarded. However, there are methods that are less sensitive to such potential bias. In this project we will estimate the effect of obesity on risk of cancer by Mendelian randomization (MR), to determine a more unbiased causal estimate for obesity on cancer risk. MR is an instrumental variables approach used in observational epidemiology to obtain consistent estimates of putative causal relationships. The method is applied when unmeasured or unknown confounding and/or reversed causation might bias the estimated association between exposure and outcome. We will perform sex-stratified, two-sample MR analysis using publicly available genome-wide association studies (GWAS) summary statistics for body mass index (BMI) and for different cancers with data downloaded from the GWAS catalog, the MR-base database or from different GWAS consortia for different cancers as in previous studies by the group. No individual level and other types of sensitive human data will be handled in the project.