This study determined the day-to-day reliability of NIRS-derived oxygenation responses (ΔStO 2%) for isometric contractions and for cuff occlusion. Twenty-four subjects (12 males and 12 females) were tested on two days (4-6 days interval). Variables generated were: (i) ΔStO 2% for isometric contractions (10%, 30%, 50% and 70% MVC) for descending trapezius (TD) and extensor carpi radialis (ECR) muscles; (ii) slope changes in total haemoglobin (HbTslope) and deoxyhaemoglobin (HHbslope) for the ECR using upper arm venous (VO, 50 mmHg) and arterial occlusion (AO, 250 mmHg); (iii) recovery slopes (Rslope) for oxygen saturation (StO2) following isometric contractions and AO. For each variable an intraclass correlation (ICC) was calculated to assess the ability to differentiate between subjects, and limits of agreement (LOA) were computed to assess day-to-day consistency of the measurement. ICCs for ΔStO2% were lowest at 10% MVC for both ECR (0.58) and TD (0.55), and highest at 30% MVC for ECR (0.95) and at 70% MVC for TD (0.79). For both muscles, LOA for ΔStO 2% was lowest at 10% and highest at 50% and 70% MVC. ICC for HbTslope was 0.17. For HHbslope ICC was higher for AO (0.83) than for VO (0.73), and LOA was lower for AO. For the ECR Rslope ICCs ranged 0.88–0.90 for contraction, but was lower for AO (0.33); LOA was lowest at 70% MVC. For trapezius Rslope ICCs ranged 0.63–0.73 and LOA was lowest at 30% MVC. For this study establishing reliability data for the ECR and TD, and including variables commonly reported, are expected to have meaning for future NIRS studies of work-related upper-extremity pain as well as for other NIRS research and clinical applications.