The range of leucocyte labelling efficiencies with 111In-oxine for a group of patients extended significantly (P = 0.05) below that obtained for a series of labellings of the same normal blood from a volunteer. A retrospective analysis was made of the results in the two groups to identify the cause of this difference in range. The labelling efficiency for patients did not vary with the volume of 111In-oxine, and was independent of the whole blood leucocyte concentration. The difference between the average labelling efficiencies obtained for a group of patients and the normal series labelled by the same operator was more significant than the difference in average labelling efficiencies obtained by different operators. It was concluded that biological variation in patients' blood, rather than operator technique, must have been a more important cause to the difference in the labelling efficiency range between patients and normal. It was also concluded that variations of contaminant platelet-bound activity and of plasma viscosity were greater in the patient group than the normal series, and contributed to this difference in labelling efficiency range.