The risk of bleeding among thrombocytopenic patients was evaluated using our new cone and platelet analyser (CPA) test. Using this test, adherence of platelets was quantitated on extracellular matrix and expressed as percent of surface coverage (SC) and the average size (AS) of aggregates. 42 thrombocytopenic patients with ITP (n=23), post chemotherapy (n= 12) and others (n= 7) were tested over a total of 82 visits. On each visit, complete blood count and CPA tests were performed and patients were evaluated for evidence of bleeding (found in 40 visits). Bleeding patients had significantly lower platelet counts (27.4 +/- 22.0 v 47.1 +/- 21.0 x 10(9)/l), lower haematocrit values (30.2 +/- 8.1 v 35.2 +/- 6.6%), lower MPV (6.83 +/-1.89 v 8.98 +/- 1.13 fl), and lower SC (4.87 +/- 3.95 v 10.33 +/-5.48%) and AS (33.99 +/- 14.94 v 52.9 +/- 24.34 microm2). Univariate analysis yielded platelet count < or =20.0 x 10(9)/l, MPV < or =8 fl, haematocrit <35%, SC <5%, AS< or =40 microm2 as significantly associated with bleeding, whereas only MPV and SC were associated with bleeding (OR 6.95, CI 2.25-21.46 and OR 4.27, CI 1.29-14.16, respectively) by multivariate analysis. When taken together, 21/22 of patients (95%) with both low SC (<5%) and low MPV (<8.0 fl) had bleeding symptoms, whereas only 9/43 (21%) patients with both these parameters above these values experienced bleeding symptoms. We conclude that the CPA test and the parameter SC (<5%) together with MPV (< or =8 fl) might be used as independent predictors of bleeding in the management of thrombocytopenic patients.