The aim of this study was to prospectively evaluate the predictive value of (18)F-fluorodeoxyglucose-positron emission tomography (FDG-PET) to detect the absence of pathological response to preoperative chemotherapy in patients (pts) with breast cancer. 63 consecutive pts with non-metastatic, non-inflammatory breast cancer, eligible for neoadjuvant chemotherapy (3 FEC 100 followed by 3 Docetaxel) were enrolled. FDG-PET was performed just before the first as well as before the second course. Metabolic activity (tumour FDG uptake) was measured by standardised uptake value (SUV(max)). Pts were classified as non-responders (NR) when the decrease of SUV(max) in the primary tumour was less than 15% at the time of the second PET (EORTC 1999 criteria). The metabolic response in FDG-PET was correlated with WHO criteria (clinical evaluation and ultrasound and/or mammography) evaluated after three cycles, pathological complete response (pCR) after surgery (according to Sataloff classification) and 4-year relapse-free survival (RFS). The mean SUV(max) decrease according to histological response was -52 ± 21% in case of pCR (Sataloff A) and 25 ± 34% in other cases (Sataloff B + C + D). Out of the 16 pts with no PET response (SUV decrease less than 15%), only one had a clinical response after the third cycle, and no pCR was observed. The 4-year RFS rate was significantly longer for metabolic responders than for NR (respectively, 85 vs. 44%; P = 0.01). This prospective study shows that a decrease in the SUV of less than 15% after the first chemotherapy course is a very potent predictor for failure of neoadjuvant chemotherapy, especially of pCR. It is interesting to note that this was shown despite the fact that the chemotherapy regimen was changed after the third course.