The CALUX (chemically activated luciferase expression) bioassay based on rat hepatoma (H4IIE) cells is a sensitive assay for the detection of Ah receptor agonists like 2,3,7,8-substituted chlorinated dibenzo-p-dioxins and dibenzofurans and related PCBs. In this paper, the assay was optimized and applied for monitoring levels of dioxins in human milk samples. Combination effects of dioxin-like compounds were evaluated by testing potential mechanisms of interaction between seven of the major dioxin-like compounds in human milk using the isobole method. Results showed that the compounds acted additively, indicating that the usual assumption of additivity in the risk assessment process is valid. In general the relative potencies (REPs) of the single agents were in accordance with their TEFs assigned by the World Health Organisation, except for the mono-ortho-substituted PCB118 that had a 40-fold lower REP in CALUX. The total dioxin-like activity was determined in 16 Danish human milk samples and was in the range 20.5-55.8 pg TEQ g(-1) fat. These values were compared with TEQs obtained from GC/MS analysis (range 14.8-43.6 pg TEQ-g(-1) fat) that overall were a little lower than CALUX TEQs. The results obtained with the bioassay when testing milk extracts fractionated into dioxins/furans, non-ortho PCB and mono/di-ortho PCB fractions indicated that the correlation between the bioassay and the chemical analyses depends primarily on the Ah receptor activity observed in the mono/di-ortho PCB fraction.