Levels of polybrominated diphenyl ethers (PBDEs) are increasing in the environment and may cause long-term environmental problems. Developing a model describing the chemical variation among the 209 possible congeners would be a useful step in any systematic approach for assessing the fate and risk posed by the PBDEs. Therefore, 40 physicochemical descriptors were derived for all PBDEs using a semi-empirical method (AM1), molecular mechanics, and empirically estimated parameters. Descriptors included heats of formation, frontier molecular orbital energies, atomic charges, dipole moments, log P values, and molecular surface areas. Principal component analysis (PCA) was used to evaluate the descriptors. The first four PCs, explaining 76% of the variation in the data, described the size, charge distribution and symmetric elements of the congeners. A quantitative structure–activity relationship model was constructed based on data for dioxin-like activity (using the luciferase bioassay) for 17 PBDEs with the partial least squares method. In addition, quantitative structure–property relationship models for gas chromatographic relative retention times on four capillary columns were developed. These models proved suitable to assist in the identification of untested PBDEs. Based on the results of the PCA, a factorial design was applied for selecting 21 representative congeners, PBDEs 11, 13, 17, 32, 35, 47, 53, 77, 85, 99, 119, 135, 153, 155, 156, 169, 176, 181, 190, 192, and 209. The spacing of these congeners in the physicochemical domain maximises the coverage of key factors such as molecular size and substitution pattern. Consideration of the selected congeners should be useful for guiding the synthesis of new compounds for use in future studies of the fate and biological effects of PBDEs.