A 4-year study was conducted to investigate the suitability of the Tradescantia-micronucleus bioassay (Trad-MCN) to detect genotoxicity of emissions from municipal waste incinerators. The genotoxic potential of emissions was assessed in three ways: (i) direct fumigation of Tradescantia with diluted incinerator emissions, (ii) exposure of Tradescantia to smoke condensates and (iii) in situ monitoring of genotoxicity of emissions near a municipal incinerator. The results of direct fumigation with smoke emissions varied over the course of the study. In 1992, significantly higher micronuclei frequencies (MCN) were always observed in treated plants than in control plants, regardless of length of fumigation period. Furthermore, MCN frequencies were significantly higher in plants fumigated for 24 h than those fumigated for 8 h, and MCN frequencies were also higher in plants fumigated in chamber 1 (low dilution, high stack gas concentration) than those in chamber 2 (P<0.05). MCN frequencies were substantially lower in 1993 and 1994, and in only 1 out of the 5 fumigation experiments were MCN frequencies significantly higher in treated plants than in controls. This drop in MCN frequencies was possibly due to the installation of better pollution control devices in the incinerator in 1993. Exposure to smoke condensates induced significant increases in MCN frequencies in most exposure trials. A drop in MCN frequencies over the course of the 4-year study was also attributed to the installation of new smoke scrubbing devices. This was further confirmed by the chemical analysis of the condensate, which showed a significant reduction in organic pollutants after the installation of the new devices. The results of in situ monitoring of genotoxicity near a municipal waste incinerator showed that MCN frequencies of Tradescantia depended on the distance from the incinerator, and the direction of wind. Because exposure periods of 6 h were insufficient to elicit significant responses, periods of 10 to 24 h are recommended for in situ monitoring.