The phosphatidylinositol 3-kinase (PI3K) pathway is frequently activated in human cancers by gain-of-function mutations of phosphoinositide-3-kinase, catalytic, alpha polypeptide (PIK3CA) or dysfunction of phosphatase and tensin homolog deleted on chromosome 10 (PTEN). Therefore PI3K is thought to be a promising target for cancer therapy. Many agents targeting PI3K have been developed and some of them have been evaluated in clinical trials. In recent years, development of predictive biomarkers as companion diagnostics for molecular targeted drugs has become an important requirement for clinical development; however, no clinically established biomarkers that predict the efficacy of PI3K inhibitors have been found. We previously reported that expression of phosphorylated Akt determined by immunoblot analysis correlated with the antitumor efficacy of a PI3K inhibitor ZSTK474 in vitro and in vivo, suggesting that it might be used as a predictive biomarker. In this study, to evaluate biomarker candidates in in vivo tumor samples, we developed an immunohistochemical protein detection/quantification system in conjunction with the tissue microarray technology using a panel of 24 human tumor xenografts (JFCR24). We have clearly demonstrated that expression levels of phosphorylated v-akt murine thymoma viral oncogene homolog (Akt) and mitogen-activated protein kinase (MAPK) determined by this system significantly correlated with those determined by immunoblot analysis. As expected, PTEN status correlated with expression of phosphorylated Akt but not MAPK. Finally, we confirmed that phosphorylated Akt levels determined using this system correlated with the in vivo efficacy of ZSTK474. The present results indicate that the immunohistochemical protein detection/quantification system could be used to quantify expression of biomarker proteins in xenografted tumor tissues as well as in human tumor specimens to predict drug efficacy in future clinical trials.