A well-differentiated colorectal tumor T 219 which grows as a xenograft in athymic mice (human-tumor-nude-mouse system) and forms colonies in culture (soft agar colony-formation assay) has been used to test the correlation between the above two methods of exposure of human tumor cells to antineoplastic agents. In in vitro studies, two protocols were used: 1 h drug exposure and continuous drug exposure. In the 1 h drug exposure experiments six drugs, doxorubicin (DX), 4'-deoxydoxorubicin (deoDX), 4'epidoxorubicin(epiDX), 4'-O-methyldoxorubicin (O-DX), N-trifluoroacetyldoxorubicin-14-valerate (AD-32) and 5-fluorouracil (FUra) were studied, while in continuous drug exposure experiments four of the above drugs (DX, deoDX, epiDX, O-DX) were studied. The survival of the tumor clonogenic cells (HC219) was determined by counting the number of colonies formed during 13-14 days of incubation and dose-response curves were obtained. In in vivo studies, the mice were treated with all of the drugs used in in vitro 1 h drug exposure experiments (DX, deoDX, epiDX, O-DX, AD-32 and FUra). To quantitate the chemotherapeutic effectiveness of the drugs, T/C% (relative tumor volume of treated group as percentage of the control group) values were calculated each time the tumors were measured. The experimental data suggest that in vitro 1 h drug exposure results are in good agreement with the in vivo results, while the continuous drug exposure results do not agree with the in vivo data. The most active drug in in vivo studies, deoDX, was found to be the most active drug in the in vitro 1 h drug exposure experiments as well. However, in continuous drug exposure experiments, O-DX, not deoDX, was found to be the most active drug. Activities of the other drugs tested also differed from their respective activities in in vivo studies. Although the relative effectiveness of various drugs can be compared by determining molar concentrations of the drugs producing 50% inhibition of colonies (ID50) the expression, PEI = LD10/ID50 X 1000, which takes into consideration toxicity of the drugs, is probably a better indicator of the in vitro drug activity. The results suggest that soft agar colony-formation assay (with established cell lines from the same tumor) may be used for the prediction of in vivo activity of potential antineoplastic agents against human tumor xenografts in nude mice.