Estimating the fertility of a semen sample or of the male from where it has been collected by simple assessment of in vitro sperm characteristics is still difficult, owing to the variable correlations that laboratory results achieve with in vivo fertility. A major reason behind these variations is the fact that the ejaculate and the artificial insemination (AI)-doses it generates are composed of a diverse sperm population. Such heterogeneity is reflected both in differences of intactness of attributes needed for fertilization, such as motility or morphology, but also in the relative ability of spermatozoa to prevail fertile over time, handling and exposure to different stimuli, all of which account for innate variations in fertilizing ability among doses, ejaculates and sires. However, methods are already available to select sub-populations of intact spermatozoa which can be tested for their degree of competence for fertilization and whose estimative power is promising, allowing the elimination of cases of sub-fertility, particularly in bovine. Examples of these methods are the separation of viable spermatozoa by swim-up or discontinuous gradient centrifugation, followed by testing the ability of the selected spermatozoa to dose-response/time sustain capacitation and acrosome reaction induction. Finding how large a sperm population with non-compensable attributes for fertilization and ability to display and sustain stimuli is, perhaps by a quick screening of membrane integrity and stability by multi-parametric methods, would allow, provided the particular male produces this sub-population in a repeatable manner, for a better estimation of fertility.