Treatment of patients with cancer-associated venous thromboembolism (VTE) remains a major challenge. The modified Ottawa score is a clinical prediction rule evaluating the risk of VTE recurrences during the first six months of anticoagulant treatment in patients with cancer-related VTE. We aimed to validate the Ottawa score using data from the RIETE registry. A total of 11,123 cancer patients with VTE were included in the analysis. According to modified Ottawa score, 2,343 (21 %) were categorised at low risk for VTE recurrences, 4,525 (41 %) at intermediate risk, and 4,255 (38 %) at high risk. Overall, 477 episodes of VTE recurrences were recorded during the course of anticoagulant therapy, with an incidence rate for low, intermediate, and high risk groups of 6.88 % (95 % CI 5.31-8.77), 11.8 % (95 % CI 10.1-13.6), and 21.3 % (95 % CI 18.8-24.1) patient-years, respectively. Overall mortality had an incidence rate of 21.1 % (95 % CI 18.2-24.3), 79.4 % (95 % CI: 74.9-84.1), and 134.7 % (95 % CI: 128.3-141.4) patient-years, respectively. The accuracy and discriminating power of the modified Ottawa score for VTE recurrence was modest, with low sensitivity, specificity and positive predictive value, and a C-statistics of 0.58 (95 % CI: 0.56-0.61). In our analysis, the modified Ottawa score did not accurately predict VTE recurrence among patients with cancer-associated thrombosis, thus hindering its use in clinical practice. It is time to define a new score including other clinical predictors.