The phosphatidylinositol 3-kinase (PI3K) pathway is activated in many human tumors and mediates processes such as cell proliferation, survival, adhesion, and motility. The natural product, wortmannin, has been widely used to study the functional consequences of PI3K inhibition in both normal and transformed cells in culture but is not a suitable cancer chemotherapeutic agent due to stability and toxicity issues. PX-866, an improved wortmannin analogue, displays significant antitumor activity in xenograft models. Here, we directly compare PX-866 and wortmannin in human cancer cell lines cultured in monolayer or as three-dimensional spheroids. Both PI3K inhibitors failed to inhibit monolayer cell growth at concentrations up to 100 nmol/L but strongly suppressed spheroid growth at low nanomolar concentrations, with PX-866 showing greater potency than wortmannin. Relative to wortmannin, PX-866 treatment results in a more sustained loss of Akt phosphorylation, suggesting that the increased potency of PX-866 is related to a more durable inhibition of PI3K signaling. PX-866 and wortmannin both inhibit spheroid growth without causing cytotoxicity, similar to known cytostatic agents, such as rapamycin. PX-866 also inhibits cancer cell motility at subnanomolar concentrations. These findings suggest that the antitumor activities of PX-866 stem from prolonged inhibition of the PI3K pathway and inhibition of cell motility. In addition, we propose that the use of three-dimensional tumor models is more predictive of in vivo growth inhibition by PI3K inhibitors in cancer cell lines lacking phosphatase and tensin homologue activity or expression.