Objective: Venous thromboembolism (VTE) may be a marker of occult cancer in the general population. While liver disease is known to increase the risk of VTE and cancer, it is unclear whether VTE in patients with liver disease is also a marker of occult cancer. Design: A population-based cohort study. Setting: Denmark. Participants: We used population-based health registries to identify all patients with liver disease in Denmark with a first-Time diagnosis of VTE (including superficial or deep venous thrombosis and pulmonary embolism) during 1980-2010. Patients with noncirrhotic liver disease and patients with liver cirrhosis were followed as two separate cohorts from the date of their VTE. Measures: For each cohort, we computed the absolute and relative risk (standardised incidence ratio; SIR) of cancer after VTE. Results: During the study period, 1867 patients with non-cirrhotic liver disease and 888 with liver cirrhosis were diagnosed with incident VTE. In the first year following VTE, the absolute risk of cancer was 2.7% among patients with non-cirrhotic liver disease and 4.3% among those with liver cirrhosis. The SIR for the first 90 days of follow-up was 9.96 (95% CI 6.85 to 13.99) among patients with non-cirrhotic liver disease and 13.11 (95% CI 8.31 to 19.67) among patients with liver cirrhosis. After 1 year of follow-up, SIRs declined, but remained elevated in patients with non-cirrhotic liver disease (SIR=1.50, 95% CI 1.23 to 1.81) and patients with liver cirrhosis (SIR=1.95, 95% CI 1.45 to 2.57). Conclusions: VTE may be a marker of occult cancer in patients with liver disease.