Background: I-SPY is a multi-center trial designed to identify predictive markers of pathological complete response (pCR) and survival of women with locally advanced breast cancers (3cm or greater). Women received neoadjuvant doxorubicin and cyclophosphamide then paclitaxel. Methods: 237 women enrolled, 216 completed serial imaging and core biopsies. Pre-treatment assays include: Agilent expression arrays, MIP aCGH, p53 gene chip and sequencing, IHC and reverse phase protein arrays (RPMA). Response to therapy was measured by serial MRI, pCR and residual cancer burden (RCB). Associations among molecular markers, pCR, RCB and survival were evaluated using chi-square test, Kaplan-Meier curves and log-rank test. Results: Median tumor size was 6cm, % pCR and RCB 0/1 was 27% and 36% for the entire study; % pCR rate for the 144 Agilent arrays was 25%. Distribution, rates of pCR and RCB 0/1 are shown in the Table for molecular and IHC markers. DFS and OS will be presented. Several molecular subtypes, including NKI 70 gene low, luminal A, 21 gene set low and IHC HR+, define 15–28% of patients with 3–10% pCR, yet excellent early survival. Wound healing, most discriminatory for prognosis, is not predictive of chemotherapy response. By RPMA, patients with pCR had increased phosphorylation of 4EBP1, eNOS, cAbl, STAT5, EGFR, AKT (p<0.05). In ER+ patients with poor MR response, pIRS, pIGFR, p706S were activated (p<0.05). RCB is a more refined way to measure pCR and was more predictive of DFS and OS (p=0.01) than pCR alone with a mean follow up of 3.9 years. MR volume is highly predictive of pCR and RCB. For specific subtypes, e.g. basal, RCB is predictive of DFS (p<0.00001). Conclusions: LABC have aggressive biology. Response to therapy and outcome can be predicted by many biomarkers. The I-SPY data set provides a platform to compare, contrast and combine marker signatures to tailor therapy and demonstrates the power of the neoadjuvant setting. Support: ACRIN U01 CA079778; CALGB CA31964, CA33601; NCI SPORE CA58207.