Water content reflectometers allow temporal and continuous assessment of spatial differences in soil water dynamics. We hypothesized that volumetric soil water content estimated by the water content reflectometers (CS616 Campbell Sci. Inc., Logan, UT) is influenced by clay content and temperature and therefore site-and or soil-specific equations are required for accurate estimations of soil water. Objectives of the study were to develop calibration equations and to evaluate soil water dynamics for an agroforestry system using the improved calibration equation. Putnam silt loam (fine, smectitic, mesic Vertic Albaqualfs) and Menfro silt loam (fine-silty, mixed, superactive, mesic Typic Hapludalfs) soils were selected with 23-54% clay. Soils were packed in cylinders and sensors were monitored at 5, 10, 15, 20, 25, 30, 35, and 40°C. Calibration equations for volumetric water content (h v ) as a function of sensor measured period, temperature, and clay content were developed. Coefficient of determination (r 2 ) and root mean square error (RMSE) were used to compare goodness of fit. RMSE varied between 0.028 and 0.040 m 3 m -3 for soil specific and soil-combined linear and quadratic equations with period. Coefficients of determination ranged between 0.89 and 0.96 for these calibrations. RMSE decreased and r 2 increased as temperature was included. The effect of temperature varied with water content, with the strongest effect at high water contents. Clay content did not contribute significantly to improve predictability. Water content estimated by the linear calibration equation with period and temperature showed differences in h v influenced by vegetation and soil depth, and closely followed precipitation events and water use by vegetation. The field study showed significant differences between the two treatments. Also the importance of temperature correction is emphasized during periods with large diurnal fluctuations and site specific calibration equations. Results of the study showed that water content reflectometers can be used to estimate h v with less than ±4% error and may need site specific calibration and a temperature correction to research more precise estimates.