Chick embryo mesoderm cells are various stages of differentiation were cultured in three-dimensional matrices of hydrated collagen. The tissues used were: stage 5 mesoderm from regions adjacent to the primitive streak; stage 12 mesoderm, comprising somitic, unsegmented (segmental plate) and lateral plate mesoderm; and stage 18 sclerotome. Explants were examined by phase contrast microscopy, including time-lapse, and scanning and transmission electron microscopy. The cells showed an increased ability to adhere to, and move in, the collagen gel with advancing stage. Of the stage 12 tissues, the unsegmented mesoderm was initially the slowest to grow out of the explant. Sclerotome cells showed by far the greatest ability to move within the gel. Where the collagen fibrils were randomly oriented, the cell morphology was polypodial and advancing lamellipodia showed clear undulations at their leading edges. A distinction was drawn between these undulations and the classical major ruffles which are seen in two-dimensional culture to uplift and pass back along the cell surface. The latter were not seen in the collagen matrix and were presumably suppressed by the three-dimensional culture configuration while the leading edge undulations were not. Ultrastructural examination showed that the cells possessed patches of amorphous material on their surface, which was sometimes interposed between the plasma membrane and collagen fibrils. Addition of hyaluronic acid (2 mg/ml) had an effect only the segmented mesoderm, where outgrowth was enhanced. Although the addition of plasma fibronectin (50 micrograms/ml) to the cultures did not affect any of the tissues, the removal of this substance, by antifibronectin antiserum or by the use of fibronectin depleted serum, inhibited outgrowth in most cases. The only tissue not reproducibly inhibited in this way was sclerotome. Alignment of the collagen fibres by the explants was observed, accompanied by an elongation of the outgrowing cells which, in bipolar form, preferentially moved up and down the aligned tracts. Scanning electron microscopy suggested that cell processes attached to, and presumably exerted tension on, bundles of fibrils thereby pulling them into line. Cell-to-cell contact was not accompanied by contact paralysis as judged by time-lapse micrography.