The mammalian zygote is a totipotent cell that generates all the cells of a new organism through embryonic development. However, if one asks about the totipotency of blastomeres after one or two zygotic divisions, opinions differ. As it is impossible to determine the individual developmental potency of early blastomeres in an intact embryo, experiments of blastomere isolation were conducted in various species, showing that two-cell blastomeres could give rise to a new organism when sister cells were separated. A mainstream interpretation was that each of the sister mammalian blastomeres was equally totipotent. However, reevaluation of those experiments raised some doubts about the real prevalence of cases in which this interpretation could truly be validated. We compiled experiments that tested the individual developmental potency of early mammalian blastomeres in a cell-autonomous way (i.e. excluding nuclear transfer and chimera production). We then confronted the developmental abilities with reported molecular differences between sister blastomeres. The reevaluated observations were at odds with the mainstream view: A viable two-cell embryo can already include one non-totipotent blastomere. We were, thus, led to propose a revised model for totipotency continuity based on the construction of the zygote as a mosaic, which accounts for differential inheritance of totipotency-relevant components between sister blastomeres. This takes place with no preordained mechanisms that would ensure a reproducible partition. This model, which is compatible with the body of data on regulative properties of mammalian early embryos, aims at tempering the rigid interpretation that discounted maternal constraints on totipotency.