Abstract
A review is given of several calculation methods used to obtain a statistically valid estimate of an expiration date from isothermal accelerated temperature studies. Common problems and how to handle them are described for this class of data. A novel computer generated plot to compare zero, first and second order decay is presented as is a somewhat simplified set of nonlinear equations. An example of the recommended analysis, which meets FDA guidelines, is given which shows that good estimates of the expiration date can be calculated from relatively poor data without knowing the degradation order. A Fortran nonlinear least squares estimation program for the analysis is described.