PIK3CA (phosphoinositide-3-kinase, catalytic, alpha polypeptide) mutations can help predict the antitumor activity of phosphatidylinositol-3-kinase (PI3K)/mammalian target of rapamycin (mTOR) pathway inhibitors in both preclinical and clinical settings. In light of the recent discovery of tumor-initiating cancer stem cells (CSCs) in various tumor types, we developed an in vitro CSC model from xenograft tumors established in mice from a colorectal cancer patient tumor in which the CD133+/EpCAM+ population represented tumor-initiating cells. CD133+/EpCAM+ CSCs were enriched under stem cell culture conditions and formed 3-dimensional tumor spheroids. Tumor spheroid cells exhibited CSC properties, including the capability for differentiation and self-renewal, higher tumorigenic potential and chemo-resistance. Genetic analysis using an OncoCarta™ panel revealed a PIK3CA (H1047R) mutation in these cells. Using a dual PI3K/mTOR inhibitor, PF-04691502, we then showed that blockage of the PI3K/mTOR pathway inhibited the in vitro proliferation of CSCs and in vivo xenograft tumor growth with manageable toxicity. Tumor growth inhibition in mice was accompanied by a significant reduction of phosphorylated Akt (pAKT) (S473), a well-established surrogate biomarker of PI3K/mTOR signaling pathway inhibition. Collectively, our data suggest that PF-04691502 exhibits potent anticancer activity in colorectal cancer by targeting both PIK3CA (H1047R) mutant CSCs and their derivatives. These results may assist in the clinical development of PF-04691502 for the treatment of a subpopulation of colorectal cancer patients with poor outcomes.