Neoadjuvant chemotherapy is the treatment of choice for locally-advanced breast cancer and leads to down staging and improved breast-conserving therapy (BCT) rates. While its efficacy is well established, considerably less is known about the most effective regimen. We have performed a retrospective analysis of 132 breast cancer patients who had undergone neoadjuvant chemotherapy at our institution. Patients had either received a) anthracyclines ("A", n=35), b) anthracyclines and taxanes ("AT", n=55), or c) neither of the two compounds ("NoA/T", n=42). Clinical response, pathological response and survival were evaluated in each arm. While all three regimens resulted in significant tumor regression, AT was most effective with a mean tumor shrinkage of 39% (ultrasound) and 41% (mammography) (Kruskal-Wallis, p=0.004, and p=0.027). Breast conservation was achieved in 75% by AT, in 49% by A, and in 19% by NoA/T (Kruskal-Wallis, p<0.001). The treatment groups did not differ in respect to pathological complete response (pCR) (chi2-test, p=0.068), although higher cumulative anthracycline doses were predictive of pCR in multivariate analyses (p=0.022). While the mammographic but not the ultrasound-determined tumor diameter determined whether a woman underwent BCT, only an ultrasound-determined size reduction was predictive for disease-free survival (DFS) and overall survival (OS) (log rank, p=0.0093, and p=0.044, respectively). Other parameters that affected BCT rates were age (p= 0.003), year of diagnosis (p=<0.001), presence of multifocal disease (p= 0.032) and the cumulative anthracycline dose (p= <0.001). While the combination of anthracyclines and taxanes is most effective in achieving clinical remission and BCT, the cumulative anthracycline dose appears most important for achieving pCR.