The marine-derived macrolide latrunculins A and B, from the Red Sea sponge Negombata magnifica, are known to reversibly bind actin monomers, forming 1:1 complex with G-actin, disrupting its polymerization. Latrunculins have remarkable physiological properties and widely used as biochemical markers. Nevertheless, no QSAR studies have been developed for any kind of actin disruptors. In the present study, Comparative Molecular Field Analysis (CoMFA) and Comparative Molecular Similarity Indices Analysis (CoMSIA) implemented in the SYBYL software packages were used to develop predictive 3D-QSAR models. Two alignment strategies were used to fit analyzed molecules to a suitable conformational template. By means of the SYBYL multifit alignment function, the best CoMFA and CoMSIA results presented cross-validated correlation coefficient values (q(2)) of 0.621 and 0.659, and non-cross-validated values (r(2)) of 0.938 and 0.965, respectively. Comparable to multifit-derived models, CoMFA and CoMSIA 3D-QSAR models were also derived using a molecular alignment obtained by docking latrunculin derivatives into the ATP active site of actin. In addition to q(2), the predictive ability was validated using external test set of five compounds. The results of this study suggest that the established model has a strong predictive ability and can be prospectively used in the molecular design and action mechanism analysis of this kind of cytotoxic compounds.