Green-lipped mussels (Perna viridis) were exposed to water-borne benzo[a]pyrene (B[a]P) at nominal concentrations of 0, 0.3, 3 and 30 microg l(-1) for up to 12 days, and both the relative levels of DNA strand breaks (assessed using an alkaline comet assay) and the proportion of micronucleus (MN) formation were monitored in mussel haemocytes at days 0, 1, 3, 6 and 12. The results of the comet assay indicated that an increase in the proportion of strand breaks occurred generally with increasing B[a]P concentration, but a significant decrease in the levels of DNA damage was observed after exposure for 12 days at all concentrations tested, suggesting that the patterns of changes in the levels of DNA strand breakage can be explained by the threshold dependent DNA repair theory. Moreover, the relatively slow development and recovery of the DNA damage response in mussel haemocytes in comparison with previous findings utilizing P. viridis hepatopancreas suggests that the response of DNA alteration upon exposure to B[a]P may be tissue-specific in this species. Monitoring the frequency of micronucleus development in mussel haemocytes indicated both dose- and time-response relationships within the exposure period. Furthermore, the levels of DNA strand breakage correlated well with the levels of micronucleus induction, suggesting a possible cause and effect relationship between the two damage types. We suggest that DNA strand breakage and micronucleus formation in mussel haemocytes can potentially be used as convenient biomarkers of exposure to genotoxicants in the marine environment.