Glucose, pyruvate, and lactate of perilymph (PL), blood, and cerebrospinal fluid (CSF) of unexposed and sound-exposed guinea pigs under ethyl urethane anesthesia were examined with due consideration of the principal sources of error. The animals had fasted for 15--20 h before the experiment to stabilize the blood glucose level. The metabolites were determined enzymatically by means of fluorescence measurements. It was found that the glucose levels depend not only on ingestion but also on the duration of anesthesia of the animals before sampling. The mean values of the scala tympani and scala vestibuli PL and CSF did not differ significantly, being about half those of blood or plasma immediately (10--20 min) after introducing anesthesia (Table 2). This concentration difference is in disagreement with the original ultrafiltration hypothesis of PL, suggesting a blood-PL barrier for glucose. The dependence on the duration of anesthesia and on the animals' ingestion before sampling appears to be an important cause of the differences in glucose data published in literature hitherto. No influence of anesthesia on pyruvate and lactate concentrations was observed. Data obtained on unexposed control animals (Tables 3 and 4) confirmed our earlier metabolite findings (Scheibe et al. 1976, 1981). No major changes in glucose, pyruvate, and lactate concentration of PL, blood, and CSF were detectable immediately after 1 h of exposure to wide-band noise at an intensity of 120 dB SPL. The present lactate findings confirmed our earlier exposure experiments (Scheibe et al. 1976), but they did not agree with the information given by Schnieder (1974).