The performance of the Hemalog D on routine blood samples from patients was assessed over a 6 wk period, 500 samples being studied in detail. The performance of the Hematrak was assessed on 85 routine blood samples. It was found that in our hospital population at least 58% of blood samples counted by the Hemalog D would require a blood film to be examined. Occasional patients with haematological diseases had abnormalities on the blood film but had normal Hemalog 8 and Hemalog D printouts. The Hemalog D was unable to perform an accurate differential count on the blood from a large proportion of patients with uraemia, including dialysis patients. The Hematrak was imprecise because of the small number of cells counted; it was also found to be inaccurate if blood films were not made rapidly after the blood was taken.