Abstract • A predictive model for storage lipid/water distribution coefficient was developed. • The model yields outstanding fitting performance, robustness, and predictive ability. • Hydrophobic and electrostatic interactions and molecular size dominate log Klip/w. • The model can be used in a wide application domain to predict log Klip/w values. The distribution of organic compounds in stored lipids affects their migration, transformation, bioaccumulation, and toxicity in organisms. The storage lipid/water distribution coefficient (log Klip/w) of organic chemicals, which quantitatively determines such distribution, has become a key parameter to assist their ecological security and health risk. Due to the impossibility to measure Klip/w values for a huge amount of chemicals, it is necessary to develop predictive approaches. In this work, a quantitative structure-property relationship (QSPR) model for estimating log Klip/w values of small organic compounds was constructed based on 305 experimental log Klip/w values. Quantum chemical descriptors and n-octanol/water partitioning coefficient were employed to characterize the intermolecular interactions that dominate log Klip/w values. The hydrophobic and electrostatic interactions and molecular size have been found to play important roles in governing the distribution of chemicals between lipids and aqueous phases. The regression (R2 = 0.959) and validation (Q2 = 0.960) results indicate good fitting performance and robustness of the developed model. A comparison with the predictive performance of other commercial software further proves the higher accuracy and stronger predictive ability of the developed Klip/w predictive model. Thus, it can be used to predict the Klip/w values of cycloalkanes, long-chain alkanes, halides (with fluorine, chlorine, and bromine as substituents), esters (without phosphate groups), alcohols (without methoxy groups), and aromatic compounds.