Single capacitance sensors are sensitive to soil property variability. The objectives of this study were to: (i) establish site-specific laboratory calibration equations of three single capacitance sensors (EC-20, EC-10, and ML2x) for tropical soils, and (ii) evaluate the accuracy and precision of these sensors. Intact soil cores and bulk samples, collected from the top 20 and 80 cm soil depths at five locations across the Upper Mākaha Valley watershed, were analyzed to determine their soil bulk density (ρb), total porosity (θt), particle size distribution, and electrical conductivity (EC). Laboratory calibration equations were established using soil packed columns at six water content levels (0-0.5 cm3 cm-3). Soil bulk density and θt significantly varied with sampling depths; whereas, soil clay content (CC) and EC varied with sampling locations. Variations of ρb and θt at the two depths significantly affected the EC-20 and ML2x laboratory calibration functions; however, there was no effect of these properties on calibration equation functions of EC-10. There was no significant effect of sampling locations on the laboratory calibration functions suggesting watershed-specific equations for EC-20 and ML2x for the two depths; a single watershed-specific equation was needed for EC-10 for both sampling depths. The laboratory calibration equations for all sensors were more accurate than the corresponding default equations. ML2x exhibited better precision than EC-10, followed by EC-20. We conclude that the laboratory calibration equations can mitigate the effects of varying soil properties and improve the sensors' accuracy for water content measurements. © 2011 by the authors; licensee MDPI, Basel, Switzerland.