Cytofluorometry of lymphocytes is an important technique in experimental biology and clinical medicine. One source of variability in the results with this technique stems from the difficulty of delineating cell subpopulations on a visual basis. We have evaluated the performance of a novel software method (Attractor), which introduces cluster analysis for the more precise definition of cell populations. Using 115 blood samples from patients with various immunological diseases, we compared the results obtained for 19 lymphocyte cell populations employing either the Attractor program or a conventional program (Cellquest). The analysis focused on inter-observer and inter-method variability and comparability. Inter-observer variability was significantly lower with the Attractor software, particularly when quantifying small cell populations such as activated subsets. The results obtained with both methods showed high correlation coefficients except for some cell populations that were either very small or which had to be calculated from the sum of other counts. The performance of the novel flow cytometric software is similar to software programs currently in use, but it offers an advantage for the definition of small and/or activated lymphocyte subpopulations. Moreover, the consistency of the measurements is better. A major disadvantage for statistical analysis, however, is that the Attractor program has not been adapted for non-parametric data.