Hedgehog proteins have been implicated in the control of myogenesis in the medial vertebrate somite. In the mouse, normal epaxial expression of the myogenic transcription factor gene myf5 is dependent on Sonic hedgehog. Here we examine in zebrafish the interaction between Hedgehog signals, the expression of myoD family genes, including the newly cloned zebrafish myf5, and slow myogenesis. We show that Sonic hedgehog is necessary for normal expression of both myf5 and myoD in adaxial slow muscle precursors, but not in lateral paraxial mesoderm. Expression of both genes is initiated normally in rostral presomitic mesoderm in sonic you mutants, which lack all Sonic hedgehog. Similar initiation continues during tailbud outgrowth when the cells forming caudal somites are generated. However, adaxial cells in sonic you embryos are delayed in terminal differentiation and caudal adaxial cells fail to maintain myogenic regulatory factor expression. Despite these defects, other signals are able to maintain, or reinitiate, some slow muscle development in sonic you mutants. In the cyclops mutant, the absence of floorplate-derived Tiggywinkle hedgehog and Sonic hedgehog has no discernible effect on slow adaxial myogenesis. Similarly, the absence of notochord-derived Sonic hedgehog and Echidna hedgehog in mutants lacking notochord delays, but does not prevent, adaxial slow muscle development. In contrast, removal of both Sonic hedgehog and a floorplate signal, probably Tiggywinkle hedgehog, from the embryonic midline in cyclops;sonic you double mutants essentially abolishes slow myogenesis. We conclude that several midline signals, likely to be various Hedgehogs, collaborate to maintain adaxial slow myogenesis in the zebrafish embryo. Moreover, the data demonstrate that, in the absence of this required Hedgehog signalling, expression of myf5 and myoD is insufficient to commit cells to adaxial myogenesis.