As a way of delineating different levels of cancer pain severity, we explored the relationship between numerical ratings of pain severity and ratings of pain's interference with such functions as activity, mood, and sleep. Interference measures were used as critical variable to grade pain severity. We explored the possibility that pain severity could be classified into groupings roughly comparable to mild, moderate, and severe. Our hypothesis was that mild, moderate, and severe pain would differentially impair cancer patients' function. We were able to identify boundaries among these categories of pain severity in terms of their interference with function. We also examined the extent to which cancer patients from different language and cultural groups differ in their self-reported interference as a function of pain severity level. We found optimal cutpoints that form 3 distinct levels of pain severity that can be defined on a 0-10-point numerical scale. We determined that, based on the degree of interference with cancer patients' function, ratings of 1-4 correspond to mild pain, 5-6 to moderate pain, and 7-10 to severe pain. Our analysis illustrates that the pain severity-interference relationship is non-linear. These cutpoints were the same for each of the national samples in our analysis, although there were slight differences in the specific interference items affected by pain. These cutpoints might be useful in clinical evaluation, epidemiology, and clinical trials.