A literature review was conducted using original papers published during 1964-1985 on the in vitro clastogenicity of chemical substances. Results of tests on 951 chemical substances were abstracted from over 240 reports to form the database. The evaluation of these data relied on each author's original conclusion on a positive or negative outcome. Of these 951 substances, 447 (47%) were consistently positive either with or without activation; 417 (44%) were negative in the direct test but not tested with metabolic activation systems; 4 were negative but tested only with activation; and 30 (3%) were clearly negative both with and without activation. The remaining 53 substances gave variable results when tested under different experimental protocols or in different cell types, but were positive in at least one test. Although discrepant results were found associated with some cell types, the addition of metabolic activation systems tended to eliminate such variability. No one cell appeared to be superior in response to all clastogens. For screening purposes, the choice of cell may thus depend more on the general usefulness and reliability of a cell type than on a strong response to a particular chemical. However, the use of a suitable metabolic activation system does appear to be of critical importance. The concentration at which clastogenic effects were detected varied extensively for different test substances, ranging from a minimum of 4.3 X 10(-8) to 6.9 X 10(2) mM. Possible mechanisms of action for substances active at only high levels are discussed, but no satisfactory explanation is available at this time. The relevance of tests conducted at concentrations high enough to alter significantly the osmolarity and other culture conditions is considered, and caution urged in the interpretation of test results obtained under physiologically stressful conditions. The clastogenic potential was compared quantitatively using an index of effective concentration (D20) and one which estimates the number of cells with exchange aberrations expected per mg/ml (TR) for data obtained by using a uniform protocol and cultures of Chinese hamster lung (CHL) cells. Both values were distributed over a wide range, demonstrating the variety of genotoxic potential in chemicals. In general, a substance which was active at only high concentrations produced fewer exchange-type aberrations. In vivo activity, as measured by tumourigenic effect and formation of micronuclei in bone marrow, tended to be greater for substances with a D20 below 10(-2) mg/ml and a TR value over 10(3).(ABSTRACT TRUNCATED AT 400 WORDS)