The zebrafish is a leading model for studies of vertebrate development and genetics. Its embryonic motor behaviors are easy to assess (e.g. for mutagenic screens), the embryos develop rapidly (hatching as larvae at 2 days) and are transparent, permitting calcium imaging and patch clamp recording in vivo. We review primarily the recent advances in understanding the cellular basis for the development of motor activities in the developing zebrafish. The motor activities are generated largely in the spinal cord and hindbrain. In the embryo these segmented structures possess a relatively small number of repeating sets of identifiable neurons. Many types of neurons as well as the two types of muscle cells have been classified based on their morphologies. Some of the molecular signals for cellular differentiation have been identified recently and mutations affecting cell development have been isolated. Embryonic motor behaviors appear in sequence and consist of an early period of transient spontaneous coiling contractions, followed by the emergence of twitching responses to touch, and later by the ability to swim. Coiling contractions are generated by an electrically coupled network of a subset of spinal neurons whereas a chemical (glutamatergic and glycinergic) synaptic drive underlies touch responses and swimming. Swimming becomes sustained in larvae once the neuromodulatory serotonergic system develops. These results indicate many similarities between developing zebrafish and other vertebrates in the properties of the synaptic drive underlying locomotion. Therefore, the zebrafish is a useful preparation for gaining new insights into the development of the neural control of vertebrate locomotion. As the types of neurons, transmitters, receptors and channels used in the locomotor network are being defined, this opens the possibility of combining cellular neurophysiology with forward and reverse molecular genetics to understand the principles of locomotor network assembly and function.