A statistical approach using sequentially principal component analysis (PCA), clustering, and discriminant analyses was developed to identify sperm morphometric subpopulations in well-defined portions of the fresh boar ejaculate. Semen was obtained as 2 portions (the first 10 mL of the sperm-rich fraction and the rest of the ejaculate, respectively) and frozen using a conventional protocol. Before freezing, an aliquot was used for computer-assisted sperm morphometry analysis (ASMA). Postthaw quality was evaluated using computer-assisted sperm analysis (CASA), and an annexin-V/PI assay evaluated sperm membranes. The PCA revealed that 3 variables represented more than 78% of the cumulative variance in sperm subpopulations. The clustering and discriminant analyses, based on 5780 individual spermatozoa, revealed the existence of 4 sperm subpopulations. The relative percentage of these subpopulations varied between boar and ejaculate portions. Linear regression models based on measured morphometric characteristics could account for up to 36% of the percentage of intact sperm membranes postthaw. The ASMA protocol used in our study was useful to detect subtle morphometric differences between spermatozoa, and the combination of this analysis with a multivariate statistical procedure gave new information on the biological characteristics of boar ejaculates that is not given by conventional sperm analysis.