Because of concern over possible failure of the SIRTF cryogenic focus mechanism in space, the SIRTF Project Office has directed that the focus should be set before launch so that the telescope arrives in orbit as close to optimum focus as possible. Then focus evaluation and determination of any required focus change to achieve best focus must be carried out without the conventional approach of a focus slew. For these tasks we have created two methods: Simfit and Focus Diversity. Simfit is a procedure for comparing an observed stellar image with a family of simulated point-source images with a range of focus settings. With a sufficiently accurate as-built telescope model for creating the simulated images, the focus offset and direction can be accurately and unambiguously determined because of the change in image appearance with defocus. Focus diversity takes advantage of the variation of best-focus setting over the instrument's focal plane due to focal plane curvature and tilt and offsets between different instrument channels. By plotting an image quality parameter, such as noise-pixels, for observed stars at several positions on the focal plane versus a defocus variable, the focus error and direction can be determined. We have developed an efficient program for carrying out these procedures. The validity of this program has been successfully confirmed using point-source images observed with three bands of the IRAC camera during a double-pass optical test of SIRTF in a Ball Aerospace cryogenic test chamber. The two procedures are described and are illustrated with these results