Sensitivity of five human small cell lung cancer cell lines to doxorubicin was assessed by a double layer agar technique using two different bottom-layers. Neither of the bottom-layers provided proportionality between numbers of cells plated and numbers of colonies, but they were correlated by a logarithmic function. Even after correction for lack of proportionality the two assay systems provided significantly different dose-response curves. The stability of the chemosensitivity was tested after 25-30 weeks continuous in vitro culture or prolonged storage in liquid nitrogen. One cell line underwent significant changes after continuous in vitro culture whereas the cell lines tested after prolonged storage in liquid nitrogen showed only minor changes. It is concluded that instead of considering the concentration necessary to achieve a certain degree of cell kill (e.g. ID50) in one experiment on one cell line, dose-response curves obtained on several cell lines in different assay systems should be used in the evaluation of new drugs.