By improving spatial and anatomical specificity, localized spectroscopy can enhance the power and accuracy of the quantitative analysis of cellular metabolism and bioenergetics. Localized and nonlocalized dynamic (31) P magnetic resonance spectroscopy using a surface coil was compared during aerobic exercise and recovery of human calf muscle. For localization, a short echo time single-voxel magnetic resonance spectroscopy sequence with adiabatic refocusing (semi-LASER) was applied, enabling the quantification of phosphocreatine, inorganic phosphate, and pH value in a single muscle (medial gastrocnemius) in single shots (T(R) = 6 s). All measurements were performed in a 7 T whole body scanner with a nonmagnetic ergometer. From a series of equal exercise bouts we conclude that: (a) with localization, measured phosphocreatine declines in exercise to a lower value (79 ‘ 7\% cf. 53 ‘ 10\%, P = 0.002), (b) phosphocreatine recovery shows shorter half time (t(1/2) = 34 ‘ 7 s cf. t(1/2) = 42 ‘ 7 s, nonsignificant) and initial postexercise phosphocreatine resynthesis rate is significantly higher (32 ‘ 5 mM/min cf. 17 ‘ 4 mM/min, P = 0.001) and (c) in contrast to nonlocalized (31) P magnetic resonance spectroscopy, no splitting of the inorganic phosphate peak is observed during exercise or recovery, just an increase in line width during exercise. This confirms the absence of contaminating signals originating from weaker-exercising muscle, while an observed inorganic phosphate line broadening most probably reflects variations across fibers in a single muscle. Magn Reson Med, 2012. ‘ 2012 Wiley Periodicals, Inc.