The modified Ottawa score (MOS) predicted venous thromboembolism (VTE) recurrence in a cohort of patients with cancer-associated thrombosis mainly managed on an outpatient basis. We aimed to assess the prognostic value of the MOS in hospitalized patients with cancer-associated thrombosis. In 383 hospitalized patients with cancer-associated VTE from the SWIss VTE Registry, 98 (25%) were classified as low risk, 175 (46%) as intermediate risk, and 110 (29%) as high risk for VTE recurrence based on the MOS. Clinical end points were recurrent VTE, fatal VTE, major bleeding, and overall mortality at 90 days. Overall, 179 (47%) patients were female, 172 (45%) had metastatic disease, and 72 (19%) prior VTE. The primary site of cancer was lung in 48 (13%) patients and breast in 43 (11%). According to the MOS, the rate of VTE recurrence was 4.1% for low, 6.3% intermediate, and 5.5% high risk (p = 0.75); the rate of fatal VTE was 0.8, 1.9, and 2.0% (p = 0.69); the rate of major bleeding was 3.1, 4.1, and 3.6% (p = 0.92); and the rate of death was 6.1, 12.0, and 28.2% (p < 0.001), respectively. None of the MOS items was associated with VTE recurrence: female gender hazard ratio (HR) 1.26 (95% confidence interval [CI], 0.53-2.96), lung cancer HR 1.17 (95% CI, 0.35-3.98), prior VTE HR 0.44 (95% CI, 0.10-1.91), breast cancer HR 0.83 (95% CI, 0.19-3.58), and absence of metastases HR 0.74 (95% CI, 0.31-1.74). In hospitalized patients with cancer-associated VTE, the MOS failed to predict VTE recurrence at 3 months but was associated with early mortality.