A reproducible method has been developed for the negative staining of myosin molecules. The dimensions of stained molecules are in close agreement with those obtained by metal shadowing. Sharp bends in the tail, indicative of hinge regions, were observed at two positions 44 nm and 76 nm from the head-tail junction. The tail was often ill-defined at the position of the first (44 nm) bend. The bend positions may be sites of proteolytic cleavage that result in the production of long and short myosin subfragment S2. About half the molecules exhibited bending to various degrees at one or both of these positions, but cases where the tail folded back on itself in a 180 degrees bend were comparatively rare (approximately equal to 10%). However, in the absence of EGTA, a large fraction of the molecules (approximately equal to 80%) exhibited 180 degrees bends. A small region, approximately 20 nm long, at the tip of the tail often appears to be significantly different from the rest. The heads are about 19 nm long and roughly pear-shaped. Although sometimes straight, more often they show a pronounced curvature. Both senses of curvature were observed, but those curved in a clockwise manner were the most common, indicating preferential binding of one side of the head to the carbon substrate. An analysis of the different combinations of head shapes in individual molecules indicates that each head can rotate independently around its long axis. No preferred angle of orientation between the two heads in a molecule, or between either head and the tail could be found. Substructure has been observed within the heads.