The effects of chemotherapy on living tumor tissue in hamsters and rats were investigated by measuring the 31P nuclear magnetic resonance spectra using topical magnetic resonance. Human neuroblastoma, human glioblastoma, and rat glioma tumor cells were inoculated s.c. in the lumbar region of the animals. After the diameter of the tumors increased to 1.5 cm, in vivo 31P nuclear magnetic resonance spectra were measured selectively in the tumors with a TMR-32 spectrometer. Adenosine triphosphate, inorganic phosphate (Pi), phosphodiester, and phosphomonoester peaks were observed. The phosphocreatine peak was hardly detectable, adenosine triphosphate and phosphomonoester peaks were high, and tissue pH, calculated from the chemical shift of Pi, declined. Regardless of the tumor origin or the histological type, the spectral pattern of each neuroectodermal tumor was found to be essentially the same. After i.v. injection of a large dose of a chemotherapeutic agent, adenosine triphosphate peaks decreased and Pi increased gradually, resulting in a dominant Pi peak pattern after 6 to 12 hours. However, during the same period, there were no observable changes in the spectra of normal organs. These findings indicated that the drugs have a selective and direct action on the energy metabolism of tumor cells. With lower drug doses, no remarkable changes were seen in the spectrum. Measurement of in vivo 31P nuclear magnetic resonance spectra is valuable not only to investigate the energy metabolism in tumor tissue but also to evaluate the effects of chemotherapy.