Purpose Circulating tissue factor (TF) has been studied as a biomarker for predicting initial, but not recurrent, venous thromboembolism (VTE) in cancer, a setting in which predictors are incompletely understood. We evaluated the association of TF, clinical risk factors, and other biomarkers measured at the time of initial VTE with recurrent VTE in a prespecified analysis of the CATCH (Comparison of Acute Treatments in Cancer Hemostasis) trial. Methods CATCH was a randomized, multicenter trial that investigated tinzaparin 175 IU/kg once daily or dose-adjusted warfarin for 6 months in patients with cancer and acute, symptomatic VTE. TF ELISA, soluble P-selectin, d-dimer, FVIII, and C-reactive protein were assayed. Fisher's exact test was used to screen for association with VTE; competing risk regression analysis of time to recurrent VTE was conducted, accounting for multiple variables. Results The study population comprised 900 patients (recurrent VTE, n = 76; 8.4%). Of these patients, 805 had samples available for TF assay. Mean and median TF levels were 72.5 pg/mL and 50.3 pg/mL, respectively (range, 15.6 pg/mL to 4,798 pg/mL). Patients in the highest quartile of TF experienced the greatest VTE recurrence (> 64.6 pg/mL; 38 [19%] of 203 patients v 34 [6%] of 602 patients; relative risk, 3.3; 95% CI, 2.1 to 5.1; P < .001). In competing risk regression analysis of time to recurrent VTE, TF remained strongly associated with recurrent VTE (subdistribution hazard ratio [SHR], 3.3; 95% CI, 1.7 to 6.4). Other significant variables included venous compression from mass (SHR, 3.1; 95% CI, 1.4 to 6.5) and hepatobiliary cancer (SHR, 5.5; 95% CI, 2.3 to 13.6). Conclusion This is the first report, to our knowledge, to describe TF as a potential biomarker of recurrent VTE in patients with cancer who are on anticoagulation treatment. A risk-adapted strategy could help identify high-risk patients who may benefit from more intensive anticoagulation approaches.