The relatively consistent fates of the blastomeres of the frog embryo could result from (i) predetermination of the blastomeres or (ii) reproducible morphogenetic cell movements. In some species, the mixing of the cells during development provides a test between these alternative hypotheses. If blastomeres are predetermined, then random intermixing of the descendants with neighboring cells could not alter their fate. To follow cell mixing during Xenopus development, fluorescent dextran lineage tracers were microinjected into identified blastomeres at the 16-cell stage. The labelled descendants of the injected blastomeres were followed over several stages of embryogenesis. After gastrulation, the labelled descendants formed relatively coherent groups in characteristic regions of the embryo. By larval stages, most of the labelled descendants were still located in characteristic regions. However, coherence was less pronounced and individual descendants were located in many regions of the embryo. Hence, cell mixing is a slow, but progressive, process throughout Xenopus development. This is in sharp contrast to the extensive mixing that occurs during the early development of other vertebrates, such as zebrafish and mice. The slow cell mixing in Xenopus development suggests a simple mechanism for the consistent fates of cleavage-stage blastomeres. The stereotyped cell movements of embryogenesis redistribute the largely coherent descendants to characteristic locations in the embryo. The small amount of mixing that does occur would result in variable locations of a small proportion of the descendants; this could contribute to the observed variability of the blastomere fate map. Because cell mixing during Xenopus development is insufficient to challenge possible lineage restrictions, additional experiments must be performed to establish when and if lineage restrictions occur.