Differential diagnosis of thrombocytosis is not always obvious. The routine clinical chemistry laboratory classically provides only limited help in distinguishing between reactive thrombocytosis (RT) and autonomous thrombocytosis, where platelet production escapes normal regulatory processes, and which is seen in myeloproliferative diseases (MPD) such as essential thrombocythemia and polycythemia vera. We explored the clinical use of platelet distribution width (PDW) in the differential diagnosis of thrombocytosis. During a 3-month period, 250 patients presenting with a platelet count > 500 x 10(9)/L were studied; 174 were classified as having RT, 42 had a diagnosis of MPD, and 34 patients were excluded because they had a hemopathy different from MPD, and either did or did not present a known etiologic factor for RT. First, we determined that in the RT group the value of PDW was closely linked to both mean platelet volume (MPV) and platelet count (PLT) (PDW = 79.5-0.005 PLT -3.5 MPV; r = 0.848, R2 = 0.720). Therefore a new parameter, PDWresidual was defined (PDWresidual = PDWobserved -PDWexpected). Second, the discrimination between reactive and autonomous thrombocytosis obtained with PDWresidual was compared with that obtained with either PDW, MPV, or PLT. PDWresidual provided much more powerful than each of the other parameters used separately: 76% of MPD patients had a PDWresidual above the 95th percentile value of the RT population and none of the MPD patients had a PDWresidual below the 50th percentile. Thus, the combined interpretation of PLT, MPV, and PDW through the use of a PDWresidual appears highly useful in the differential diagnosis of thrombocytosis. Also, through simple modeling, more information can be drawn from parameters such as PDW that hitherto were mostly discarded as being without clinical interest.