The term 'chromosome mutations' was chosen and defined for this review to refer to alterations of chromosome structure (reciprocal, heritable translocations), of chromosome number (loss or gain of a whole chromosome), or of chromosome content (loss or gain of a part of a chromosome). Chromosome mutations may result from chromosome breakage (clastogenesis) and its consequences or from disruption of chromosome behavior during cell division (nondisjunction). State-of-the-art protocols are outlined to test for heritable translocations, for whole-or partial chromosome loss (clastogenesis), and for whole chromosome loss or gain (nondisjunction). The literature up to 1980 was reviewed and 106 papers were selected for the evaluation of 116 chemicals for one or more chromosome mutation end points. The criteria used for acceptance of data from the literature were not stringent, as most of this work was done some time ago and for purposes other than testing. The main criterion was that germ cell stage sampling was correct. For the evaluation of the accepted data, numerical requirements were set up, using as a guide the control data from all the papers. Compounds were classified, when possible, as mutagenic (+) or nonmutagenic (-). Those not classifiable, usually due to insufficient numbers of chromosomes tested, were listed as inconclusive (inc). Of 61 compounds tested for heritable translocations, 27 were positive, 8 were negative, and 26 were inconclusive. Of the 35 with conclusive data, only 21 also have definitive carcinogenesis classifications (all positive). Of these, 19 were deemed mutagenic, which gives agreement of 90.5%. Of the 76 compounds tested for clastogenesis by the chromosome loss test, 26 were positive, 13 were negative, and 37 were inconclusive. Of the 39 with conclusive data, only 20 also have definitive carcinogenesis classifications. 15 of the 19 carcinogens were positive. Four of the carcinogens were negative and 1 noncarcinogen was positive, for an overall agreement of 75%. Of 44 compounds tested for nondisjunction, 15 were positive, 13 were negative, 16 were inconclusive. Of the 28 compounds with conclusive data, only 9 have definitive carcinogenesis classifications (all positive). Five of these were deemed negative and agreement was only 44%. It should be noted that these data do not fairly represent these short-term tests as conducted with current protocols. A more equitable comparison could be achieved with planned experiments that include the sex-linked recessive lethal (SLRL) test in the comparison.