We previously reported primary endpoints of two consecutive phase I/II trials, evaluating different schedules of neoadjuvant epirubicin (E), gemcitabine (G) and docetaxel (Doc) for primary breast cancer (PBC). Here, we report mature survival data and prognostic factors. One hundred fifty-one patients were recruited into two consecutive phase I/II trials of neoadjuvant chemotherapy for T2–4 N0–2 M0 PBC. Patients received six cycles of G/E/Doc every 3 weeks with G repeated on d8 (GEDoc, n = 84) or five cycles of G/E followed by four cycles of Doc all given every two weeks (GEsDoc, n = 67). Prognostic factors were investigated using univariate and multivariate analyses. No survival differences by treatment were found. Among reported predictive factors for pathologic complete response (pCR), oestrogen receptor (ER) status was the only relevant factor in the multivariate analysis. Unexpectedly, pCR resulted in poorer survival (univariate HR for overall survival [OS] 3.11, p = 0.007). Multivariate analyses identified molecular subtype and tumour size as the most relevant prognostic factors for OS. HER2-receptor status and the CPS-EG score (Mittendorf et al., J Clin Oncol 2011;29:1956–62), based on clinical and pathological stage, ER-status and tumour grade, were particularly relevant in disease-free survival. Our findings cast doubt on the reliability of pCR as single marker for prognosis of this unselected breast cancer cohort, with an abundance of luminal subtypes. These results underline the significance of additional molecular characteristics for breast cancer survival.