Bioconcentration factors (BCFs) are of great importance for ecological risk assessment of organic chemicals. In this study, a quantitative structure-activity relationship (QSAR) model for fish BCFs of 8 groups of compounds was developed employing partial least squares (PLS) regression, based on linear solvation energy relationship (LSER) theory and theoretical molecular structural descriptors. The guidelines for development and validation of QSAR models proposed by the Organization for Economic Co-operation and Development (OECD) were followed. The model results show that the main factors governing logBCF are Connolly molecular area (CMA), average molecular polarizability (α) and molecular weight (M W ). Thus molecular size plays a critical role in affecting the bioconcentration of organic pollutants in fish. For the established model, the multiple correlation coefficient square (R 2 Y ) = 0.868, the root mean square error (RMSE) = 0.553 log units, and the leave-many-out cross-validated Q 2 CUM = 0.860, indicating its good goodness-of-fit and robustness. The model predictivity was evaluated by external validation, with the external explained variance (Q 2 EXT ) = 0.755 and RMSE = 0.647 log units. Moreover, the applicability domain of the developed model was assessed and visualized by the Williams plot. The developed QSAR model can be used to predict fish logBCF for organic chemicals within the application domain.
BCFs, QSAR, organic pollutants, applicability domainBioconcentration factor (BCF) is an important ecotoxicological parameter charactering the bioaccumulation of chemicals in organisms [1] . It is commonly used to assess the hazard and risk of new and existing substances, such as persistent organic pollutants (POPs) [2] and persistent, bioaccumulative and toxic (PBT) substances [3,4] . Although several guidelines for the experimental determination of BCF were documented [5,6] , the experimental determination of BCF values is expensive, time-consuming and complicated. Moreover, the experimental data are not available for all chemicals in use and forthcoming use. Many researchers developed BCF predictive methods by the methodology of quantitative structure-activity relationships (QSARs) [7][8][9][10][11][12][13][14][15][16] . In environmental science, QSARs are mathematical models to explore the inherent relations between molecular structures of chemicals and their physicochemical properties, environmental behavioral and ecotoxicological parameters [17] . Thus, QSARs can fill in the data gap of organic pollutants, decrease experimental expenses and especially reduce animal testing [17] .