Strong evidences suggest that many different sperm subpopulations co-exist within the mammalian ejaculate. These subpopulations have been identified in a number of species; however, to the best of our knowledge, no data exist regarding the existence of sperm subpopulations within the canine ejaculate. Ejaculates were obtained by masturbation from four mongrels and processed using a standard freezing protocol. Motility data were analysed before and after cryopreservation using a computer assisted sperm analysis (CASA) system ISAS. On raw data, a principal component analysis (PCA) was performed to reduce the number of motility descriptors to a few informative variables, and then a K-means cluster procedure was performed and then a regression analysis to validate the clusters obtained in the second analysis. ANOVAs and chi-squared analyses were used to compare clusters and males. PCA revealed that two principal components represented more that the 88% of the variance with eigenvalues of 3.25 and 3.02, respectively. The clustering and discriminant analysis using curvilinear velocity and linear velocity as variables revealed the existence of 11 sperm subpopulations--four of them characterized by high velocities, two by medium values and five by low velocities. After freezing-thawing, nine subpopulations were found--four of high velocities, two of medium and three of low velocities. It is concluded that freezing-thawing not only impairs sperm motility but also produces changes in the sperm subpopulation structure in the canine ejaculate, that the evaluation of the sperm structure subpopulations is a better indicator of semen quality and freezeability than the use of mean values, and that two sperm motility quality indexes can be used to resume of the variables obtained from the CASA analysis.