The ability to differentiate regional patterns of flow and metabolism between various patient populations depends upon the signal-to-noise characteristics of the data. The approach chosen for producing quantitative data will affect the detection sensitivity of a method. Methods based on mathematical models can reduce intersubject variability by accounting for factors unrelated to the physiological measure of interest, in particular, differences in the input function. However, errors in the model and in the implementation of a model-based method can increase variability compared to simpler, empirical methods. Normalization of physiological measures can significantly reduce intersubject variation; however, interpretation of normalized results can be more complex. The advantages and disadvantages of various approaches for physiological quantitation are considered.