This study introduced a 3D rod-and-frame test in a virtual environment (VERF) and evaluated its reliability and concurrent validity according to four scoring methods previously used with 2D tests. Thirty nonimpaired older adults (15 men; 15 women) aged 64-86 (M=73.6) manipulated a rod in a semi-immersive 3D virtual environment. Results suggested that the conventional 2D scoring method of absolute deviation from vertical was the most reliable method (alpha=.79; gamma=.89). This approach also had higher correlations with traditional neuropsychological measures of spatial ability than three other methods. Given the availability of a reliable and valid scoring method, the VERF may result in previously unavailable insights through the measurement of visuospatial abilities in 3D.