Studies on catheter-related central venous thrombosis (CRCVT) have been focused mainly on clinically evident CRCVT due to occlusive thrombi, underestimating therefore the actual thrombosis prevalence. This prospective study was aimed at evaluating prevalence, timing and evolution of thrombosis, and identifying involved veins and risk factors in cancer patients (pts) undergoing percutaneous subclavian central venous catheterization (CVC) for chemotherapy, parenteral nutrition or both. We enrolled 127 consecutive pts requiring partially or totally implanted central venous silastic catheters. The study protocol included peripheral phlebography (P) at day 8, 30 and every two months following CVC and/or when clinically indicated, along with peripheral and pullout P on catheter withdrawal. A quantitative scale was developed to evaluate thrombus grading in subclavian, innominate and cava veins. Age, sex, coagulation profile tumor histotype, metastases, therapy, catheter type, and catheter insertion side were also investigated. Only pts who underwent at least two P were evaluated, and chi 2 test was adopted for statistical analysis. Altogether, 95 pts were evaluable. CRCVT was observed in 63/95 (66%) pts. At day 8, 30 and 105 (representing the median days in which first, second and last P were performed) CRCVT was evidenced in 64%, 65% and 66% of the pts, respectively. Thrombus grading did not differ among first, second and last P. CRCVT was symptomatic in 4/63 (6%) pts. Thrombosis prevalence was higher in subclavian (97%) with respect to innominate (60%) or cava (13%) veins (p < 0.001). Thrombosis was higher in left subclavian catheters (14/16; 87.5%) than in right ones (49/79; 62%), p < 0.01. No associations were established between CRCVT and other investigated parameters. Our data show a very high actual frequency of CRCVT in cancer pts, and emphasize that first days following CVC are at the highest risk for CRCVT development. Based on our results, a study on short-term antithrombotic prophylaxis in cancer pts requiring CVC is warranted. Finally, our data indicate that left subclavian vein catheterization represents a risk factor for CRCVT.