Six healthy volunteers provided simultaneous serum and mixed saliva specimens at various time intervals after receiving three different concentrations of trimethobenzamide HCI (total dose 800 mg) via rectal syringe on three separate occasions. Specimens were analyzed for trimethobenzamide content and the results subjected to pharmacokinetic analysis and other statistical procedures to identify characteristics of the relationship between distribution processes in these specimen types. On the basis of linear regression analysis there was a significant correlation (P less than 0.01) between saliva and serum drug content for both mean values (r = 0.94) at each sampling time and all individuals' paired data points (r = 0.72). Because of the large degree of scatter in individual pairs, the data were subjected to analysis of variance in an attempt to determine a potential source of this divergence. The relationship between serum and saliva drug concentrations was found to depend upon the time after drug administration. We conclude that a nonuniform process determines the relative degree of partitioning trimethobenzamide between serum and saliva.