During the development of the nervous system, axons grow over considerable distances to reach their appropriate targets. Information derived from a range of experimental systems suggests that a multiplicity of guidance cues govern growth cone navigation. Among these may be physical features of the environment, pathways of extracellular matrix molecules such as laminin, and distributed positional information cues on the surfaces of neuroepithelial cells. Yet some of these guidance mechanisms may act only over a short range, and it is uncertain whether any of them can specify direction. A possibility that is theoretically attractive is that axons might be directed by diffusible signals emanating from their targets. Although this idea was first proposed by Ramón y Cajal at the beginning of the century, accumulating evidence that chemotropism plays a role in neural development has only recently become compelling. Some in vivo experiments have hinted strongly at chemotropism, as when axons navigate to their target along ectopic routes. But there is only one way of unequivocally demonstrating a chemotropic response of growing neurites. This involves placing an explant containing the neurons of interest at some distance from their target tissue In a three-dimensional collagen matrix devoid of other landmarks. Within such gels it has been demonstrated that gradients of diffusible molecules can be established [T. Ebendal (1977) Cell Tissue Res.175, 439-458]. During the culture period, axons may then display direct or arcuate trajectories toward the target across the neutral matrix. If this phenomenon is observed in the presence of the target but not in the presence of control tissues, this suggests that chemotropism participates in axonal pathfinding during normal development.