Central venous access devices are often essential for the administration of chemotherapy to patients with malignancy, but its use has been associated with a number of complications, mainly thrombosis. The true incidence of upper extremity deep vein thrombosis (DVT) in this setting is difficult to estimate since there are very few studies in which DVT diagnosis was based on objective tests, but its sequelae include septic thrombophlebitis, loss of central venous access and pulmonary embolism. We performed an open, prospective study in which all cancer patients who underwent placement of a long-term Port-a-Cath (Pharmacia Deltec Inc) subclavian venous catheter were randomized to receive or not 2500 IU sc of Fragmin once daily 90 days. Venography was routinely performed 90 days after catheter insertion, or sooner if DVT symptoms had appeared. Our aims were: 1) to investigate the effectiveness of low doses of Fragmin in preventing catheter-related DVT; and 2) to try to confirm if patients with high platelet counts are at a higher risk to develop subclavian DVT, as previously suggested. On the recommendation of the Ethics Committee, patient recruitment was terminated earlier than planned: DVT developed in 1/16 patients (6%) taking Fragmin and 8/13 patients (62%) without prophylaxis (Relative Risk 6.75; 95% CI: 1.05-43.58; p = 0.002, Fisher exact test). No bleeding complications had developed. As for prediction of DVT, there was a tendency towards a higher platelet count in those patients who subsequently developed DVT, but differences failed to reach any statistical significance (286 +/- 145 vs 207 +/- 81 x 10(9)/1; p = 0.067). According to our experience, Fragmin at the dosage used proved to be both effective and safe in these patients.