The reliability, relation to whole-body maximal oxygen uptake (VO(2max)), and force-level sensitivity of (31)P MRS markers of mitochondrial function were studied in 39 normal-weight women. Following 90 s isometric plantar-flexion exercises at 45, 70 and 100% of maximum voluntary contraction, skeletal muscle mitochondrial function was determined from the phosphocreatine recovery time constant (TC(PCr)), the ADP recovery time constant (TC(ADP)), and the rate of change in PCr during the first 14 s of recovery (OxPhos). VO(2max) was measured on a treadmill. Test-retest measurements were obtained in a subset of seven women. Overall, TC(PCr), TC(ADP) and OxPhos were reproducible for all exercises (coefficients of variation = 2.3-19.3%). With increasing force-level, TC(PCr) was prolonged (29.0 +/- 8.2, 31.9 +/- 9.0 and 35.4 +/- 9.5 s), OxPhos was increased (0.159 +/- 0.081, 0.247 +/- 0.090 and 0.310 +/- 0.114), and TC(ADP) was shortened (22.4 +/- 7.9, 21.3 +/- 6.2, and 19.5 +/- 6.7; p < 0.01). All MRS markers of mitochondrial function were correlated with VO(2max) (r = 0.41-0.72; p < 0.05). These results suggest that measurements of TC(PCr), TC(ADP) and OxPhos yield reproducible results that correlate with whole-body VO(2max) and that vary in force-level sensitivity.