In two institutions at Rome and Florence we evaluated the clinical sensitivity of two Coulter STKR systems using the NCCLS standard H20-T for leucocyte differential count in a patient population with high prevalence of haematologic abnormalities. Reference ranges of normal leucocytes were obtained on 278 adult subjects. On a population of 455 patient specimens, 200 specimens (44%) were flagged by the STKR because of a distributional abnormality, and 122 (27%) because of a morphological abnormality. Percentage of subtotal agreements between the STKR and the reference manual differential count was 85.4%, with 67.5% full and 20.9% partial agreements. Eight specimens that showed a morphological abnormality with the reference manual differential count were classified as normal by the STKR, with a false normal rate of 6.6%. Analysis of the STKR performance for morphological abnormalities showed acceptable sensitivity (82.0%) and rather low specificity (71.5%), low predictive value of positive results (51.3), high predictive value of negative results (91.5%) and efficiency of 74.3%. The main problems of the STKR differential count were a high rate of false monocyte count, and the misidentification of eosinophilias and low-concentration abnormal cells.