PURPOSE: The purpose of this study was to determine, using in vivo and in vitro 1H MRS (MR spectroscopy), the characteristic biochemical metabolites related with breast cancer, and to assess the clinical usefulness and limitations of this modality. MATERIALS AND METHODS: For in vivo 1H MRS, nine patients with breast cancer and two normal volunteers were examined on a 1.5 T MR imager equipped with facilities for spectroscopy. In order to localize the breast lesion, axial and sagittal T1-weighted images and fat-suppressed T2-weighted images were obtained just prior to MRS; MR spectra were acquired at TR=3000 msec and TE=144 msec. For in vitro 1H MRS, breast tumor and adja-cent normal tissue were extracted from 13 patients with breast cancer, and in two of these, both in vivo and in vitro 1H MRS were performed. All in vitro 1H MRS specimens were immediately immersed in liquid nitrogen, and then in a preparation of perchloric acid. For quantitative analysis of the MR spectra of cancerous and normal breast tissue, the paired t-test was used (p < 0.05). RESULTS: At1H MRS in vivo, choline and two lipids were identified at 3.21 ppm, and 1.33 ppm and 0.9 ppm, re-spectively. The distinction between cancerous and normal breast tissue was based on the higher level of choline (3.21 ppm) present in the former. At 1H MRS in vitro, on the other hand, mean and standard deviation (% standard deviation) for the various metabolites in cancerous and normal breast tissue were as follows: choline, 30.195 +/- 2.448(8.108) and 22.648 +/- 1.938(8.556); trimethylamine, 3.425 +/- 0.335(9.769) and 0.640 +/- 0.066(10.325); sarcosine, 3.425 +/- 0.335(9.769) and 0.640 +/- 0.099(15.394); lactate, 16.388 +/- 1.134(6.922) and 9.715 +/- 0.385(3.965); inositol, 1.970 +/- 0.282(14.334) and 3.859 +/- 0.502(13.020); and taurine, 6.614 +/- 0.556(8.412) and 10.748 +/- 1.206(11.222). High levels of choline (p=0.026), trimethylamine (p=0.001), sarco-sine (p=0.009), and lactate (p=0.009), and lower levels of inositol (p=0.006) and taurine (p=0.008) were char-acteristic findings in cancerous as compared with normal breast tissue, with significantly different results. CONCLUSION: 1H MRS both in vitro and in vivo showed that increased choline levels were present in cancerous breast tissue, but that normal tissue does not contain choline. The presence of choline could therefore be used as a marker for malignancy in breast lesions. Information provided by in vitro 1H MRS, together with the development of in vivo 1H MRS with high field strength and high resolution, may be very useful for the diagnosis of breast cancer.