The purpose of the present study was to assess the reliability of metabolic parameters measured using 31 P magnetic resonance spectroscopy ( 31 P MRS) during two standardized rest-exerciserecovery protocols. Twelve healthy subjects performed the standardized protocols at two different intensities; i.e., a moderate intensity (MOD) repeated over a two-month period and heavy intensity (HEAVY) repeated over a year's time. Test-retest reliability was analyzed using coefficient of variation ( Key words: magnetic resonance spectroscopy; reliability; oxidative capacity; PCr kinetics; exercise Phosphorus-31 nuclear magnetic resonance spectroscopy ( 31 P-MRS) is now widely accepted as the "gold standard" method for noninvasive measurements of energy metabolism in exercising muscle. The analysis of changes in phosphorylated compounds concentrations and intracellular pH during rest-exercise-recovery protocols has provided significant advance in the understanding of underlying mechanisms of muscle energy production (1-4). Given the noninvasiveness of the method, 31 P-MRS enables repeated measurements and has been successfully used in a wide range of situations such as the assessment of therapeutic interventions in patients with metabolic disorders (5-7) and the characterization of training-induced metabolic changes (8 -11). More recently, the dynamics of phosphocreatine (PCr) has been used to evaluate the control of mitochondrial respiration in skeletal muscle (12) or the mechanism underlying the V O 2p slow component (13) typically observed during a high-intensity exercise; i.e., above the lactate threshold. In addition, the rate constant of postexercise recovery kinetics of [PCr] has been used as an index of mitochondrial function (14 -17). In addition to these traditional variables, more sophisticated analyses have been put forth in order to quantify the contribution of oxidative and anaerobic pathways to energy production in exercising muscle (2,18,19).Despite the widespread use of 31 P MRS for muscle energetics investigation, the reproducibility of the corresponding variables has been scarcely investigated. It has actually been investigated through repeated measurements performed 1 week (5) and 4 weeks apart (20). Bendahan et al. (5) reported a good reliability in metabolic parameters when investigated using analysis of variance (ANOVA). However, detection of systematic bias might have been affected by large random error between tests as suggested previously (21). observed large test-retest variability in metabolic parameters measured during steady state phases of moderate isometric plantar flexion exercise. However, kinetics parameters of recovery exhibited a small coefficient of variation (CV Ͻ 9%). In this line, van den Broek et al. (22) reported a low variability (CV Ͻ 11.5%) for the parameters describing the recovery kinetics of PCr and adenosine diphosphate (ADP) in a single subject who repeated the same exercise protocol 10 times. Based on a visual inspection of the kinetics of phosphorylated compounds and pH i...