In a series of patients with pulmonary embolism (PE) we have previously demonstrated that the risk of recurrent PE was inversely correlated to platelet count (PlC) levels. To find out whether PlC levels were also associated to a different incidence of heparin-related bleeding complications, we report our experience with 1,103 consecutive patients with venous thromboembolism (VTE) receiving full-dose heparin therapy. Six points of clinical and laboratory information were recorded on admission and then compared to the development of bleeding: the patient's age and sex; the etiology of VTE; the type of heparin used (unfractioned, UFH, vs. low-molecular-weight, LMWH), the presence or lack of PE findings on lung scan, and the PC levels on admission. Bleeding occurred in 64/1,103 patients (6%). Patients who bled were significantly older than those who did not (72 +/- 11 vs. 64 +/- 17 years; p = 0.0005). There were no significant differences in bleeding rate according to any of the risk factors that could have predisposed to VTE, but patients treated with UFH bled significantly more frequently than those on LMWH (48/636 vs. 16/467; odds ratio: 2.30; 95% confidence interval: 1.25-4.28). Finally, mean PlC levels were significantly lower at VTE diagnosis in patients who subsequently bled (227 +/- 112 vs. 262 +/- 110 x 10(9) liters-1; p = 0.01). The logistic regression analysis confirmed that all three variables were independent risk factors for bleeding complications. This is the first study to demonstrate that PlC levels (within the normal range) are inversely correlated with the risk of heparin-related bleeding. These findings may be interest not only from the point of view of pathogenesis but also clinically, as they may be used in the decision as to which VTE patients could receive heparin therapy at home.