Microtubules are the primary target for many anti-cancer drugs, the majority of which bind specifically to beta-tubulin. The existence of several beta-tubulin isotypes, coupled with their varied expression in normal and cancerous cells provides a platform upon which to construct selective chemotherapeutic agents. We have examined five prevalent human beta-tubulin isotypes and identified the colchicine-binding site as the most promising for drug design based on specificity. Using this binding site as a template, we have designed several colchicine derivatives and computationally probed them for affinity to the beta-tubulin isotypes. These compounds were synthesized and subjected to cytotoxicity assays to determine their effectiveness against several cancerous cell lines. We observed a correlation between computational-binding predictions and experimentally determined IC(50) values, demonstrating the utility of computational screening in the design of more effective colchicine derivatives. The most promising derivative exhibited an IC(50) approximately threefold lower than values previously reported for either colchicine or paclitaxel, demonstrating the utility of computational design and assessment of binding to tubulin.