Over the last three years the Virginia Highway Research Council has directed a research effort toward improving the method of determining the moisture content of soils with a nuclear gage. The first task in this research was the determination of the correct moisture calibration curve for the nuclear gages being used (Troxler Models 227 and 2401). Forty-three soils sampled from the three physiographic areas of Virginia and having optimum moisture contents ranging from 5% to 40% were tested. It was found that for the Model 227 gage two calibration curves produced better results than did one calibration curve. With one curve the standard error was 2.7 pcf (0.04 g/cc), while for two curves the errors were 1.4 and 1.7 per (0.02 and 0.03 g/cc), which are judged to be acceptable for compliance testing. From an investigation of the physical and chemical properties of the soils tested it was concluded that the separation of the two curves could be based on optimum moisture content, and that the separation value should be 21%. At this point in the study the testing of the Model 227 gage was discontinued because the Virginia Department of Highways had replaced all of these gages with the Model 2401. The Model 2401 gage requires only one calibration curve, which has a standard error of 1.7 pcf (0.03 g/cc). Using this gage and its single calibration curve, five moisture standards were developed. A zero moisture content standard was developed by using dry C-190 sand, and four standards were developed by mixing different portions of either MgSO4 7 H2O or Na3PO4 12 H2O with Ottawa silica C-190 sand. These standards had respective moisture contents of 9.3, 16.1. 21.0, and 25.8 pcf (0.15, .26, .34 and .41 g/cc). The curve of best fit for these standards had a standard error of only 0.57 pcf (0.01 g/cc). Also the curve was approximately parallel to the curve previously determined from the soils. At approximately 25 pcf (0.40 g/cc) moisture content, the curve from the standards gives a moisture content about 2 pcf (0.03 g/cc) higher than the curve from the soils; and at around 5 pcf (0.08 g/cc) moisture content, the former gives a moisture content about 1 pcf (0.02 g/cc) higher than does the latter. The manufacturer's curve generally gives a lower moisture content than does either of the other two curves at the same count ratio.