1. Distributions of dense bodies per platelet were evaluated in unstimulated human platelets and in platelets stimulated to secrete by the addition of varying amounts of thrombin or the ionophore A23187. Since distributions were not Gaussian, they were compared by non-parametric methods. 2. Most distribution obtained following varying levels of dense-body release differed significantly from the control (unstimulated) distribution. In general, distribution from points widely separated in terms of per cent release of total dense-body content also differed significantly. Distributions obtained at points where the total number of dense bodies lost was approximately the same did not differ significantly, regardless of the drug and incubation conditions used to produce the dense-body loss. 3. Dense bodies were lose from platelets during the secretory process (measured as the release of either calcium or ATP), and the number of dense bodies lost correlated at a significant level with the amounts of both substances secreted. 4. Following maximal secretion produced by either thrombin or A23187, a population of dense bodies (refractory dense bodies) remained in platelets. The refractory dense bodies could not have been distributed uniformly among the control populations of either platelets or dense bodies. The distribution of refractory dense bodies could, however, be described by a model in which platelets containing large numbers of dense bodies had a greater proportion of their dense-body complement refractory than did platelets containing fewer dense bodies.