The potential of total choline (tCho) signal-to-noise ratio (SNR) (ChoSNR) and tumor volume in the assessment of tumor response in locally advanced breast cancer (LABC) patients (n = 30) undergoing neoadjuvant chemotherapy (NACT) was investigated using magnetic resonance spectroscopic imaging (MRSI) and conventional MRI at 1.5 T. Experiments were carried out sequentially at four time-points: prior to therapy and after I, II and III NACT and ChoSNR, and the tumor volume was measured. The MR response was compared with the clinical response. Sequential data of 25 patients were retrospectively analyzed by classifying them as clinical responders and non-responders. In 14 responders, the pre-therapy ChoSNR was 7.8 ± 5.1. In 10/14 responders, no choline was observed after III NACT while in the remaining four patients the ChoSNR was reduced to 3.6 ± 1.1 (p < 0.05). Non-responders showed no statistically significant change in ChoSNR. After III NACT, the tumor volume reduced by 84.0 ± 14.8% in responders. Using receiver operating curve (ROC) analysis, cut-off values of 53% for ChoSNR and 47.5% for volume were obtained to differentiate responders from non-responders. The sensitivity to detect responders from non-responders using ChoSNR was 85.7% with 91% specificity while 100% sensitivity was observed for volume but with reduced specificity of 73%. Our results indicate that ChoSNR may serve as a useful parameter to predict tumor response to NACT with higher specificity compared to volume, suggesting its potential in effective treatment management. Copyright © 2010 John Wiley & Sons, Ltd.