risk of several clinical features, including malignancy. The process factors out of the equation apparent predictors that are no more than covariants of more powerfully contributing risk factors. The generalizability from these studies is limited by their small size and relative lack of clinical validation, but it is of interest that none of these formulas includes malignancy among its final contributors, whereas age and body weight consistently appear (20). This makes it probable that the excess risk of DVT following cancer surgery is due not so much to the presence of malignancy itself, as to an accumulation of other wellknown and commonly associated risk factors which include advanced d5a, debility, prolonged and difficult surgery, and a lengthy and complicated post-operative course, all of which add to the onset of DVT in their own right. For instance, deep leg vein thrombosis (DVT) after colo-rectal surgery is linked much more closely to the onset of infectious complications than with the presence of cancer (21). A similar approach has been taken to the identification of very high risk patients among those having cancer surgery. Postoperative DVT in 17.5 % of 4ll patients with gynaecological malignancy wits significantly associated by single variate analysis with recurrent malignancy, leg oedema, the presence of varicose veins, non-white genetic background, a past history of DVT, previous radiation therapy, extensive surgery, &go and body weight, the duration of anaesthesia and surgical blood loss; but logistic regression analysis reduced the number of contributing variables to the duration of anaesthesia, uEa, previous DVT , tace, leg oedema and varicose veins, in that order of importance (22). An added element which has been too little explored is the likelihood that DVT in cancer patients may develop during the time leading up to surgery, so that peri-operative prophylaxis comes too late.