Phosphorus nuclear magnetic resonance (31P NMR) was used to study energy metabolism in the rat mandibular gland. The gland was isolated, perfused arterially and set in the NMR tube. At rest, 7 resonance peaks were observed and 6 peaks identified from low field as: 1) sugar phosphates (SP) and nucleotide monophosphate (NMP), 2) inorganic phosphate (Pi), 3) creatine phosphate (PCr), 4) gamma-nucleotide triphosphate (NTP) and beta-nucleotide diphosphate (NDP), 5) alpha-NTP, alpha-NDP, NAD+, and NADH, 6) an unknown peak, and 7) beta-NTP. From the results of high performance liquid chromatography (HPLC), NTP consisted mainly of ATP and GTP, and UTP was not detected. The tissue contents of ATP and GTP in the perfused gland were determined by HPLC as 1.86 +/- 0.03 and 0.37 +/- 0.01 mmol/kg wet tissue (S.E., n = 5). From 31P NMR and HPLC data, the tissue levels of creatine phosphate, ADP, and sugar phosphates were estimated as 3.3, 0.4, and 4.2 mmol/kg wet tissue, respectively. The cessation of perfusion decreased the tissue levels of PCr and ATP and increased those of Pi and SP. On the other hand, administration of acetylcholine (1 microM), which is an optimal dose for secretion, decreased PCr and increased Pi but did not change SP. The ATP was unchanged initially and slowly decreased to the lower level during sustained secretion. These findings suggest that a sustained secretion requires more energy from ATP hydrolysis rather than initial secretion.