A study was conducted at the Los Alamos National Laboratory (LANL) to evaluate the performance of the SNAP (Spectral Nondestructive Assay Platform) analytical software for measurements of known standards in large metal waste boxes (2.5 m3 volume). The trials were designed to test the accuracy and variance of the analytical results for low-density combustible matrices and higher-density metal matrices at two discrete gamma-ray energies: 121.78 keV and 411.12 keV. For both matrix types the measurement method that produced the most accurate results with the lowest associated standard deviation involved combining four individual measurements taken at the geometric center of each of the box's four vertical sides. With this method the overall bias and the standard deviation amongst 24 individual results for the 121.78 keV and 411.12 keV gamma rays were as follows: 3.38% (+ or = 20.19%) and 3.68% (+ or = 15.47%) for the combustible matrix and 37.88% (+ or = 67.64%) and 9.38% (+ or = 33.15%) for the metal matrix. The persistent positive bias from measurements of the metal box is believed to be a result of a nonhomogenously distributed matrix.