As an experimental strategy for potentially dissociating and studying the cytotoxic and cytodifferentiative antileukemic effects of 6-thioguanine (6-TG), cultured human promyelocytic leukemia cells (HL-60) were serially selected for growth in increasing concentrations of 6-TG (0.5 to 50 micrograms/ml). Three acquired characteristics, cytotoxic resistance, cytodifferentiative resistance, and double minute chromosomes (DM), were monitored at successive 6-TG selection levels. Approximately 200-fold resistance to the cytotoxic effect of 6-TG was acquired at the first selection step, and it neither increased at higher 6-TG selection levels nor reverted to greater sensitivity in cells subcultured off of drug. This was due to the irreversible loss of hypoxanthine-guanine phosphoribosyltransferase (HPRT) activity. In contrast, a lesser, not completely quantifiable, degree of resistance developed to the cytodifferentiative effects of the purine nucleobases hypoxanthine and 6-TG which varied as a function of 6-TG selection pressure. Numerous DM, not observed in the parental wild-type HL-60 cells, appeared at 6-TG (0.5 micrograms/ml) selection which varied substantially in parallel with 6-TG selection pressure up to 6-TG (20 micrograms/ml). At higher selection levels (50 micrograms/ml or prolonged culture on 20 micrograms/ml), a marked decrease in DM occurred which was associated with the acquisition of new marker chromosomes. The most consistent marker was a chromosome 6 with additional material in the short arm (6p+); this was noted as a single copy in the basal 6-TG/20 subline but as two copies (trisomy 6; 2p+) in independently selected higher 6-TG-resistant subcultures. These cytogenetic findings suggest the presence of amplified genes which increased in number and shifted from a predominance in extra-chromosomal DM to intrachromosomal sites as a function of 6-TG selection. Among the 6-TG-resistant sublines, there was no change or a decrease in the amplification level of the known amplified oncogene c-myc from that demonstrated in parental HL-60 cells. Although proof requires detailed analyses with specific gene probes, the overall results imply that: (a) the cytotoxic component of the resistance is due to an invariant loss of HPRT which, therefore, is not likely to be related to amplified genes; (b) the cytodifferentiative component of the resistance is due to a positively selectable mechanism which could be directly or indirectly related to 6-TG-selected amplified genes; and (c) variations in the cytogenetic indicators of amplified genes and the resistance to 6-TG cannot be simply ascribed to quantitative variations in c-myc amplification.