Dynamic phosphorus magnetic resonance spectroscopy (31P-MRS) with high temporal resolution enables, through measurement of phosphocreatine (PCr) recovery after exercise, a noninvasive evaluation of mitochondrial capacity of the skeletal muscle in vivo. The aim of this master thesis is the comparison of different parameters that can possibly influence the results of 31P-MRS investigation on muscle metabolism. To this end, three independent studies were performed at two MR sites. First study deals with the impact of ergometer type, coil diameter and magnetic field strength within the same group of subjects. Measurements of the calf muscle exercise in eleven subjects on a 3T MR system with home-made mechanical ergometer and on 3T and 7T MR systems equipped with commercial pneumatic ergometer were performed at three different workloads. Higher magnetic field strength improves the data quality (higher SNR, better fitting accuracy), increasing reliability, but no significant differences were found in the metabolic parameters between magnetic field strengths. In the inter-institution comparison, significant differences were found in the concentration of PCr measured at rest and in the depletion of PCr causing significant difference in mitochondrial capacity (Qmax) at low workload. From this study it can be concluded that metabolic parameters measured by dynamic 31P-MRS do not depend upon used magnetic field strength. However, same workload and same protocols for measurement and evaluation are important for comparison of data acquired with different ergometers, especially when effects of very mild exercise are examined. Second study is dedicated to evaluate the depth resolved surface coil MRS (DRESS) as a non-echo based localization method for dynamic 31P-MRS at 7T. Two dynamic examinations of plantar flexions at 25% of maximal voluntary contraction, one without spatial localization and one with the DRESS slap positioned obliquely over the gastrocnemius medialis, were conducted in fourteen volunteers. Significant differences in metabolic parameters were found between the non-localized and DRESS localized data. Splitting of the inorganic phosphate (Pi) signal was observed in several non-localized data sets consisting of a mixture of signals from several muscles, but in none of the DRESS-localized data sets. In conclusion, the DRESS localization scheme yields good spatial selection and provides muscle specific insight into oxidative metabolism even at relatively low exercise load. Third study targeted on the comparison of single voxel localization through slice selective localization by adiabatic selective refocusing (semi-LASER), with DRESS and surface coil localization. Five volunteers performed three dynamic examinations on plantar flexions, one with each localization method, at the same workload. Signal from gastrocnemius medialis acquired by DRESS and semi-LASER localization shows a higher PCr depletion, a trend towards lower tau and lower pH at the end of exercise in comparison to non-localized acquisition. Since the relatively long echo time needed for semi-LASER limits the quantification of J-coupled spin systems (e.g., ATP), a calculation of Qmax is not possible. The results suggest a favored application of semi-LASER than DRESS when higher anatomical and physiological specifity is needed with the advantage of DRESS in calculation of Qmax.