So-called “air mass functions” of photovoltaic modules are used to approximate the effects of spectral responsivity and to correct short-circuit current to or from a reference condition. These empirical functions are determined from outdoor measurements with test modules mounted on two-axis solar trackers and then calculated from plots of normalized calibration value (short-circuit current divided by total irradiance) versus optical air mass. Because they are incorporated into a number of photovoltaic system modeling and sizing software programs, the accuracy of the functions has direct implications for system costs. We discuss the assumptions associated with these functions that are generally not considered or ignored, and study their variability with respect to atmospheric constituents. The variability study included a 6-month outdoor measurement on a crystalline-Si module and a software simulation of the same module using a solar spectral irradiance model. We conclude that air mass functions depend on the measurement location and time, and therefore are not unique to a particular device. Also, using these functions introduces two distinct errors, the magnitudes of which are unknown without knowledge of spectral irradiance conditions. Published 2012. This article is a U.S. Government work and is in the public domain in the USA.