The platelet count has a primary role in the diagnosis and treatment of idiopathic thrombocytopenic purpura (ITP). This study analysed the accuracy of ITP patient platelet counts determined by Abbott CD-Sapphire (impedance/optical) and Bayer Advia 120 (optical) analyses, compared with a reference immunoplatelet method. Instrument platelet estimates showed broad equivalence in the higher range of observed values, but significant discrepancies against the immunoplatelet count were seen when platelet counts were <10 x 10(9)/l. CD-Sapphire mean platelet volume (MPV) results revealed increased (>12 fl) platelet volumes in eight of eight ITP patients with counts of <20 x 10(9)/l compared with 6/6 and 5/13 patients with platelet counts of 20-50 and >50 x 10(9)/l. In contrast, Bayer Advia MPV values showed no relationship with the platelet count. Increased reticulated platelets were associated with an increasing CD-Sapphire MPV (R(2) = 0.61) and a decreasing platelet count. High (>40%) reticulated platelet values were seen in 9/9 patients with immunoplatelet counts of <20 x 10(9)/l compared with 0/19 patients with platelet counts above 20 x 10(9)/l. There may be a need for caution in the interpretation of platelet counts in ITP patients obtained with conventional instrument methods, and therapeutic decisions should ideally be validated by reference immunoplatelet procedures.