The aim of this work was to develop a predictive model for milk/water partition coefficients of neutral organic compounds. Batch experiments were performed for 119 diverse organic chemicals in human milk and raw and processed cow milk at 37 °C. No differences (< 0.3 log units) in the partition coefficients of these types of milk were observed. The polyparameter linear free energy relationship model fit the calibration data well (SD = 0.22 log units). An experimental validation data set including hormones and hormone active compounds was predicted satisfactorily by the model. An alternative modelling approach based on log K_ow revealed a poorer performance. The model presented here provides a significant improvement in predicting enrichment of potentially hazardous chemicals in milk. In combination with physiologically based pharmacokinetic modelling this improvement in the estimation of milk/water partitioning coefficients may allow a better risk assessment for a wide range of neutral organic chemicals.