Particular non-monotonic dose-response curves of many endocrine disrupting chemicals (EDCs) suggest the existence of diverse toxicity mechanisms at different dose levels. As a result, the biological activities of EDCs cannot be simply exhibited by unique EC50/LD50 values, and the quantitative structure-activity relationship (QSAR) analysis for non-monotonic dose-response relationship becomes an unknown field in the environmental science. In this paper, nine phenols with inverted U-shaped dose-response curves in lymphocyte proliferation test of Carassius auratus were selected. The binding interactions between the phenols and several typical EDCs-related receptors were then explored in a molecular simulation study. The estrogen receptor (ER), androgen receptor (AR), thyroid hormone receptor (TR), bacterial O2 sensing FixL protein (FixL), aryl hydrocarbon receptor (AhR), and the peroxisome proliferator-activated receptor (PPAR) were the target receptors in the study. Linear regression QSAR models for the low and high exposure levels of the compounds were developed separately. The results indicated that the lymphocyte proliferation in the low-dose range might involve ER–mediated process, while the proliferation inhibition in the high dose range was dominated by the acute toxicity of phenols due to receptor occupancy and cell damage. Particular non-monotonic dose-response curves of many endocrine disrupting chemicals (EDCs) suggest the existence of diverse toxicity mechanisms at different dose levels. As a result, the biological activities of EDCs can not be simply promoted by unique EC50 / LD50 values, and the quantitative structure-activity relationship (QSAR) analysis for non-monotonic dose-response relationship becomes an unknown field in the environmental science. In this paper, nine phenols with inverted U-shaped dose-response curves in lymphocyte proliferation test of Carassius auratus were selected The binding interactions between the phenols and several typical EDCs-related receptors were explored in a molecular simulation study. The estrogen receptor (ER), androgen receptor (AR), thyroid hormone receptor (TR), bacterial O2 sensing FixL protein ), aryl hydrocarbon receptor (AhR), and the peroxisome proliferator-activated receptor (PPAR) were the target receptors in the study. Linear regression QSA R models for the low and high exposure levels of the compounds were developed separately. The results indicated that the lymphocyte proliferation in the low-dose range might involve ER-mediated process, while the proliferation inhibition in the high dose range was dominated by the acute toxicity of phenols due to receptor occupancy and cell damage.