The objective of this study was to establish in vivo (1)H-magnetic resonance (MR) spectroscopic appearances of cervical cancer using an endovaginal receiver coil and corroborate findings with magic angle spinning (MAS) MR spectroscopy of tissue samples. Fifty-three women (14 controls and 39 with cervical cancer) underwent endovaginal coil MR imaging at 1.5 T with T(1)- and T(2)-weighted scans sagittal and transverse to the cervix. Localized (1)H MR spectra (PRESS technique, TR 1600 ms, TE 135 ms) were accumulated in all controls and 29 cancer patients whose tumour filled > 50% of a single 3.4 cm(3) voxel. Peaks from triglyceride-CH(2) and -CH(3) were defined as present and in-phase (with the choline resonance), present but out-of-phase, or not present. Peak areas of choline-containing compounds were standardized to the area of unsuppressed tissue water resonance. Comparisons in observed resonances between groups were made using Fisher's exact test (qualitative data) and a t-test (quantitative data). Biopsies from these women analysed using MAS-MR spectroscopy and normalized to the intensity of an external standard of silicone rubber were similarly compared. Adequate water suppression permitted spectral analysis in 11 controls and 27 cancer patients. In-phase triglyceride-CH(2) resonances (1.3 ppm) were observed in 74% of tumours but in no control women (p < 0.001). No differences were observed in the presence of a 2 ppm resonance, choline-containing compounds or creatine in cancer compared with control women. However, ex vivo analysis showed significant differences not only in -CH(2), but also in -CH(3), a 2 ppm resonance, choline-containing compounds and creatine between tissues from control women and cancer tissue (p < 0.001, = 0.001, = 0.036, < 0.001 and = 0.004 respectively). On in vivo (1)H-MR spectroscopy, the presence of positive triglyceride-CH(2) resonances can be used to detect and confirm the presence of cervical cancer. However, technical improvements are required before routine clinical use.