Résumé : The purpose of this study is to compare Near Infrared Spectroscopy (NIRS) thenar eminence parameters obtained with 2 different devices from the same manufacturer (InSpectra Models 325 and 650, Hutchinson Tech, Min USA), and 2 different probes (15 vs. 25 mm spacing), in healthy volunteers (HV) and ICU patients. Prospective, observational study in ICU setting. Simultaneous, cross over NIRS inter-device comparison and comparison between different probes (25 vs. 15 mm spacing) were done at baseline and during vascular occlusion tests (VOTs). Forty patients (19 septic shock, 21 trauma), and 29 HV were included. NIRS inter-device comparison showed similar baseline StO2 values in HV and patients. The VOT result were significantly different for minimal StO2 value reached during VOT (StO2min) (intraclass concordance coefficient (ICC) = 0.18), the occlusion slope (ICC = 0.16) and the reperfusion slope (StO2reperf) (ICC = 0.26). The probe comparison was also significantly different for VOT parameters (StO2min (ICC = 0.43), occlusion (ICC = 0.50) and StO2reperf (ICC = 0.48). The low concordance, poor agreement and large bias (ICC and Bland & Altman) observed, were related both to the device used and the probe spacing. StO 2 data obtained with NIRS model 650 and 15 mm probe differ from values obtained with the previous device (325 and probe spacing 25 or 15 mm). This difference is not related to the population tested, but to the device and probe spacing. As a consequence, despite similar trends for variations between HV and patients during VOT, threshold and predictive values for outcome should be revisited with the new device before the acceptance for routine clinical use. © 2012 Springer Science+Business Media New York.