Near-infrared spectroscopy (NIRS) and laser Doppler flowmetry (LDF) have become the techniques of choice allowing the non-invasive study of local human skeletal muscle metabolism and blood perfusion on a small tissue volume (a few cm3). However, it has been shown that both NIRS and LDF measurements may show a large spatial variability depending on the position of the optodes over the investigated muscle. This variability may be due to local morphologic and/or metabolic characteristics of the muscle and makes the data interpretation and comparison difficult. In the present work, we use a third method to investigate this problem which permits fast, non-invasive mapping of the intramuscular vessel distribution in the human vastus latelralis muscle. This method uses an advanced, passive, infrared imaging sensor called a QWIP (quantum well infrared photodetector). We demonstrate, using a recovery-enhanced infrared imaging technique, that there is a significant presence of perforator vessels in the region of interest of approximately 30 x 18 cm (the number of vessels being: 14, 9, 8, 33, 17 and 18 for each subject, respectively). The presence of these vessels makes the skeletal muscle highly inhomogeneous, and may explain the observed NIRS and LDF spatial variability. We conclude that accurate comparison of the metabolic activity of two different muscle regions is not possible without reliable maps of vascular 'singularities' such as the perforator vessels, and that the QWIP-based imaging system is one method to obtain this information.