Schizophrenia, bipolar disorder and major depression have been reported to be associated with some cancers. However, the magnitude of the causal relationship between them remains unclear. This study aims to explore the potential association between three major mental diseases and the risk of some cancers. We performed the two-sample Mendelian randomization(MR) analysis using publicly available genome-wide association studies (GWAS) statistics to investigate the causal relationship between these three mental diseases and some common types of cancers, including breast cancer, ovarian cancer, lung cancer, colorectal cancer, bladder cancer, prostate cancer, thyroid cancer, pancreatic cancer, malignant melanoma and glioma. We obtained genetic association estimates for schizophrenia, bipolar disorder and depression from the Psychiatric Genomics Consortium.The genetic association estimates for cancers were obtained from the UK Biobank, the MRC-IEU consortium and the GliomaScan consortium. After correction for heterogeneity and horizontal pleiotropy, we detected suggestive evidence for the association between thyroid cancer and genetically predicted schizophrenia (OR = 1.543, 95% CI: 1.023–2.328, P = 0.039), and thyroid cancer and major depression (OR = 3.573, 95% CI: 1.068–11.953, P = 0.039). No evidence of causal effects of schizophrenia, major depression and bipolar disorder on other types of cancers. Our findings suggest the association of schizophrenia and major depression and the development of thyroid cancer. •Mental illness is closely related to various types of cancer.•Mendelian randomization is a novel and persuasive causal reasoning method.•Association between schizophrenia, bipolar disorder, major depression and cancer•Genetic link between schizophrenia, major depression and thyroid cancer