Developing blood vessels were observed directly on the dorsal surface of the optic tectum of anesthetized, transparent albino Xenopus laevis tadpoles, stages 41-54. Case histories of individual tadpoles indicated that pial capillaries developed by the classical mechanism of sprouting of endothelial cells from existing blood vessels. 'Deep sources' appeared on the tectal surface during development. These were sites of upwelling blood cells from capillaries within the nervous tissue of the tectum into vessels on the surface. Few 'deep sinks' were observed in the dorsal tectum of normal tadpoles. The earliest deep sources were probably formed by sprouts from the surface vessels through the basement membrane and into the nervous tissue; later ones may also have formed from internal sprouts back to the surface. Maps of deep sources and of surface vessels in case histories indicated that neural tissue and blood vessels in the caudal half of the tectum grew faster than in the rostral half. The medial venules on the dorsal tectum originated as ordinary-sized rostrocaudal capillaries. They enlarged in diameter as they drained the increasing flow of blood from the tectum into the choroid plexus over the 4th ventricle. Some capillaries disappeared or regressed during development. Our observations on the tectum were consistent with the classical sequence of loss of flow, narrowing, collapse of the lumen, and retraction of endothelial cells into adjacent vessels. Likely sites for regression were upstream from a deep source and at crosslinks between transverse vessels on the lateral tectum. Morphometric parameters for tectal angiogenesis were (a) surface density (mm-1) calculated as total length of surface vessels divided by the dorsally projected surface area, and (b) density of deep sources (mm-2) calculated as total number divided by surface area. From stages 41/42 to 50 average surface density approximately doubled, and average density of deep sources increased about 5-fold. Some of the factors which might be expected to alter brain angiogenesis include nervous activity, availability of O2, and metabolic rate. Removal of one eye deprived the contralateral tectum of direct retinal inputs, while the ipsilateral side was a control in the same animal. Anterograde labeling of retinal axons with diI18 from the remaining eye confirmed projections only to the opposite side. No significant differences in densities of surface vessels or of deep sources were observed between the contralateral and ipsilateral sides of the tectum.(ABSTRACT TRUNCATED AT 400 WORDS)