The need for quality control of leucoreduction of blood products has led to the development of various methods to count low levels of residual leucocytes. We compared five platforms side-by-side: the Nageotte haemocytometer and four based on fluorescent staining of nuclei: two flowcytometers (Beckman Coulter, BD Biosciences) with methods based on counting beads, a volumetric flow cytometer (Partec) and the microvolumic fluorimeter ImagN2000 (BD Biosciences), all according to their manufacturers' recommended methods. Analysis of double-filtered red cell concentrates (RCCs) and platelet concentrates (PCs), spiked with various numbers of leucocytes, revealed good linearity for all methods over the range of 1.6-32.7 leucocytes/microl, all with r(2) > 0.99. At the rejection level of leucocyte-reduced blood components, i.e. 1 x 10(6) per unit corresponding with approximately 3.3 leucocytes/microl, the Nageotte haemocytometer had low accuracy (0% for RCCs, 56% for PCs), and was relatively imprecise [coefficient of variance (CV) of 34% and 30% respectively]. The Partec flow cytometer gave good results for RCCs (accuracy 67%, CV 22%), but not for PCs (accuracy 0%, CV 25%). The ImagN2000 had an accuracy of 44% for RCCs and 89% for PCs, but the precision was variable (CV 32% for RCCs, 15% for PCs). The best results were obtained with the Beckman Coulter (RCCs: accuracy 86%, CV 13%, PCs: accuracy 67%, CV 16%), and BD Biosciences platforms (RCCs: accuracy 100%, CV 10%; PCs: accuracy 89%, CV 11%). We conclude that, at the rejection level of 1 x 10(6) leucocytes per unit, the widely used Nageotte haemocytometer performs poorly in terms of inaccuracy and imprecision, and that both counting-bead-based, flow cytometric methods performed best.