The effects of four gravel concentrations (0, 20, 40, and 60 percent by weight) and three gravel size fractions (4–8 mm, 8–15 mm, and 15–40 mm) on the calibration of a neutron moisture meter and density probe was investigated for a beach sand and a clayey tropical Alfisol (Oxic Paleustalf). The slope of the regression line of the calibration curve relating neutron count ratio with volumetric moisture content of gravel-free sand was statistically significantly different from that of three gravel-sand mixtures. The calibration curve of three gravel-sand mixtures was statistically identical. For the gravel-soil mixtures, the slope of the calibration curve generally decreased with increased gravel concentration. The statistical comparison of the regression lines for the 4–8 mm gravel size fraction indicated that the slope coefficient of the calibration curve of only 60 percent gravel-soil mixture was significantly lower than that of 0, 20, and 40 gravel percentages. For the 8–15 mm gravel size fraction, the slopes of the regression lines of 40 and 60 percent gravel-soil mixtures were similar and significantly lower than that of 0 and 20 percent gravel concentrations. Gravel size had no effect on the calibration curve in the gravel-sand mixtures. In the gravel-soil mixtures, for 20 and 40 percent gravel concentrations, 4–8 mm gravel size had a significantly higher slope coefficient of the regression equations, compared with the 8–15 mm size fraction. The slope of the regression equation relating density count ratio with dry density significantly decreased with an increase in gravel concentration for 40 and 60 percent, compared with 0 and 20 percent gravels for gravel-soil mixtures. Heterogeneity of tropical soils, with regard to variations in texture and gravel concentrations, may necessitate a detailed calibration of the neutron moisture meter prior to its use on large watersheds.