The hippocampal formation is a neuroanatomically well-defined region of the brain involved in memory processes. In view of the functional importance of the region and its involvement in a number of brain pathologies, including Alzheimer's disease and temporal lobe epilepsy, a quantitative description of its vascular supply represents an important first step in evaluating the involvement of vascular changes in these phenomena. Unbiased estimates of the length and connectivity of the vascular supply of brain regions have not been described previously. The total number, total length, and distribution of the diameters of capillaries were estimated in the five major subdivisions of the hippocampal formation (fascia dentata, hilus, CA3-2, CA1, and subiculum) in 5 normal males, 52-84 years of age. These estimates were used to derive several other structural parameters. Both the primary and the derived parameters were used to make inter- and intra-individual comparisons. For each of the five major subdivisions from each individual, the volume was estimated using the Cavalieri principle. The total capillary length was estimated on 3-microm-thick plastic isotropic uniform random sections. Using a topological definition of a capillary unit and the optical disector, total capillary number was estimated in 40-microm-thick plastic sections. Length-and number-weighted three-dimensional diameter distributions were obtained from the thin and thick plastic sections, respectively. In each subdivision the total length of capillaries was correlated with previously obtained data on the number of neurons in the same subdivisions of the same individuals. Intersubdivisional differences were observed, in that the hilus of the dentate gyrus had fewer capillaries per unit volume than the other four subdivisions. Interindividual comparisons indicate that the interindividual variances are of a magnitude suitable for sensitive group comparisons. The design-based stereological methods that were used in the analyses can provide a basis for a new unbiased approach to the estimation of vascular parameters in well-defined regions of the brain.