Regulatory agencies governing the disposal of nuclear waste require that the waste be appropriately characterized prior to disposition. The most important aspect of the characterization process, establishing radionuclide content, is often achieved by nondestructive assay (NDA). For NDA systems to be approved for use in these applications, measurement uncertainty must be established. Standard propagation of errors methods provide a good starting point for considering the uncertainty analysis of NDA systems for nuclear waste. However, as compared with other applications (e.g., nuclear material accountability), using NDA systems for nuclear waste measurements presents some unique challenges. These challenges, stemming primarily from the diverse nature of the waste materials encountered, carry over into the uncertainty analysis as well. This paper reviews performance measures appropriate for the assessment of NDA uncertainty, describes characteristics of nuclear waste measurements that contribute to difficulties in assessing uncertainty, and outlines some statistics based methods for incorporating variability in waste characteristics in an uncertainty analysis.