The selection of the starting dose for initial clinical trials of anticancer agents is mostly determined by toxicological endpoints in mice (LD10). So far, very few attempts have been made to evaluate the potential value of cytotoxicity assays for this purpose. The present study was undertaken as a first attempt to investigate the relationship between cytotoxicity of anticancer drugs in vitro and pharmacokinetic parameters in vivo in patients, at suggested maximum tolerated doses. Using the fluorometric microculture cytotoxicity assay (FMCA), we determined the concentration giving 50% cell survival (IC50) in vitro, for 25 cytotoxic drugs in fresh preparations of normal peripheral blood mononuclear cells (PBMC) and of tumor cells from patients with acute or chronic lymphocytic leukemia (ALL or CLL). Using linear regression, we investigated the relationship between the IC50s and clinically achievable peak plasma concentrations (Cmax) or concentration-time products (C x T) in humans. The clinical data was obtained from the literature. Based on all drugs tested, good correlations were obtained between IC50s for CLL cells, and both Cmax and C x T (R approximately 0.7, p < 0.0002), and for ALL cells and normal PBMC between IC50 and Cmax, while the two latter cell types showed somewhat weaker relationships to C x T. Using the IC50 data of CLL cells, predictions of Cmax and C x T exceeded 1 log for only four drugs. No tendencies to under- or overprediction within different classes of drugs were noted. The results demonstrate a significant relationship between toxicity in vitro and achievable systemic exposure of anticancer drugs in humans, which suggests that non-clonogenic in vitro assays for drug sensitivity testing may provide pharmacokinetic information useful in the development of investigational cytotoxic drugs.