One of the problems of near-infrared-spectroscopy (NIRS) measurements is low reproducibility. The aim of the present study was to introduce quality criteria to increase reproducibility of peripheral NIRS measurements. In a prospective cohort study in 40 neonates, repeated NIRS measurements were performed on the calf. During five "reapplication" periods (of NIRS optodes), five "measurements" (venous occlusions) were performed. Tissue oxygenation index (TOI), mixed venous oxygenation (SvO2), fractional oxygen extraction (FOE), hemoglobin flow (Hbflow), oxygen delivery (DO2), and oxygen consumption (VO2) were assessed. Measurements with linear changes during venous occlusions were included for further analysis (first quality criterion: R(2)>0.95). The second quality criterion was the equation 0 < or = TOI-SvO2 < or = (SaO2-SvO2)x0.2. Variance components and mean standard deviations were analyzed after introduction of the quality criteria. Variance components of reapplication and measurement decreased after introduction of the second quality criterion (TOI: 46.6-35.0%, SvO2: 76.8-38.2%, FOE: 73.1-37.5%, Hbflow: 70.3-51.9%, DO2: 71.5-52.7%, and VO2: 70.9-63.8%). Mean standard deviations of TOI (6.6+/-3.0 to 4.7+/-3.2%), SvO2 (11.1+/-4.8 to 5.7+/-3.9%), FOE (11.3+/-4.8 to 5.9+/-4.0%), Hbflow (4.3+/-2.0 to 2.9+/-1.6 micromol100 mLmin), and DO2 (17.8+/-7.6 to 11.4+/-6.2 micromol100 mLmin) decreased significantly, too. Only 12% of measurements fulfilled both quality criteria. With the introduction of two quality criteria, test-retest variability of peripheral NIRS measurements decreased significantly and reproducibility increased significantly.