We investigated a series of Taxol@ (1. paclitaxel) analogues using a 3-D QSAR approach (CoMFA). Published and unpublished data for more than 50 compounds have been included in the analysis and the model has been used to predict activity of other analogues. The model accurately describes known SARs and also has good predictive power. As a guide to the design of new and more potent analogues of the anticancer drug 1, a 3-D QSAR model (CoMFA2) was developed using a training set of 49 analogues of 1 and the biological activities for binding to microtubules. Further, the biological activities of an additional 47 analogues were predicted from the model. Structures and activity data were taken from a review and from unpublished results by Georg and coworkers.