Previous studies have shown that total platelet count (TPC) inadequately predicts bleeding in thrombocytopenic patients with haematological malignancies. This prospective cohort study evaluated whether rotational thromboelastometry (ROTEM), coagulation or other platelet parameters were more strongly associated with bleeding than TPC. Adults treated at two UK haematology centres for haematological malignancy were enrolled if they had thrombocytopenia (TPC ≤ 50 × 10(9) /l) at beginning of, or during treatment (International Standard Randomized Controlled Trial Number 81226121). TPC and bleeding symptoms were recorded daily for up to 30 d or until platelet count recovery, hospital discharge or death. Blood samples were tested thrice weekly using ROTEM, Platelet Function Analyser (PFA)-100(®) , coagulation and platelet cytometry assays. Bleeding symptoms and TPC from 49/50 enrolled participants who completed the study were recorded on 754/760 study days. Mean platelet volume and PFA-100(®) closure times were frequently inestimatable because of thrombocytopenia. TPC, absolute immature platelet number (AIPN) and ROTEM maximum clot firmness were significantly associated with bleeding on the day after blood sampling. Only AIPN was associated with bleeding after adjustment of test results for TPC (Odds Ratio 0·52, 95% confidence interval 0·28-0·97; P = 0·038). In a predictive model, AIPN was superior to TPC for predicting bleeding. This study indicates that AIPN may be more clinically useful than TPC at predicting bleeding.