Recent developments in 13C nuclear magnetic resonance (NMR) spectroscopy have permitted noninvasive assessment of glycogen concentration in human skeletal muscle. Before these indirect measurements could be accepted as accurate, it was essential that validation should be carried out by comparing the widely used method of muscle biopsy and direct biochemical assay for glycogen concentration with measurement by NMR. Eight normal subjects underwent six NMR scans of gastrocnemius and three biopsies of the same muscle on the same day. The overall mean for muscle glycogen concentration was 87.4 mM by NMR and 88.3 mM by biopsy. There was a close correlation between the pairs of observations on each subject (R = 0.95; P less than 0.0001). The mean coefficient of variation for NMR measurement was 4.3 +/- 2.1% and that for biopsy was 9.3 +/- 5.9%. The performance of the muscle biopsies was accompanied by a small but significant rise in plasma-free fatty acids (529 +/- 157 to 667 +/- 250; P less than 0.01), epinephrine (17 +/- 6 to 25 +/- 8 pg/ml; P less than 0.02), and norepinephrine (318 +/- 119 to 400 +/- 140 pg/ml; P less than 0.02) but no change in plasma glucose, plasma insulin, nor muscle glycogen concentration assessed by NMR. The study demonstrates that in vivo 13C NMR measurement of human muscle glycogen can be regarded as accurate, and the technique is associated with a higher precision that biopsy with direct biochemical assessment.