Several alternative algorithms for computer-assisted derivation of measurements of movement characteristics from manually reconstructed tracks of progressively motile human spermatozoa were compared. Fifty tracks were reconstructed at 30 Hz from video recordings and analysed using traditional manual methods and by four combinations of computer algorithms. The best algorithm set was identified ('Videomot.mdpt') and the values for the curvilinear, average path and linear velocities (VCVL, VAVE and VLIN respectively), the amplitude of lateral displacement of the sperm head about the axis of progression (AH) and the number of times the sperm head crossed the average path (the 'beat/cross frequency', BXF) obtained using it were compared to those obtained by manual analysis. There was a considerable time saving when the computer-assisted method was used and the values it gave for the various movement characteristics showed good correspondence with those obtained manually. In addition, repeated data entry and analysis was found to be highly reproducible. When the tracks were re-plotted at 6 Hz (as used by the multiple-exposure photomicrography method for sperm motility analysis) insufficient information remained in the tracks for reliable determination of anything other than VLIN. We conclude that the Videomot.mdpt program provides reliable values for the movement characteristics of progressively motile human spermatozoa, and believe it will be of great value in the validation of commercial systems providing automated sperm movement analysis and in laboratories which do not have access to such costly equipment.