Prior to launch, the Spitzer Space Telescope (SST) secondary focus mechanism was set to a predicted desired in-orbit focus value. This predicted setting, determined from double-pass cold chamber measurements and calculated groundto-orbit corrections, had an uncertainty greater than the required in-orbit focus accuracy. Because of concern about the potential for failure in a cryogenic mechanism affecting all Spitzer instruments, it was required that any focus correction be made in a set of moves directly from the initial to the desired setting. The task of determining the required focus moves fell to IRAC (Infrared Array Camera), the instrument most affected by and sensitive to defocus. To determine the focus directly from examining images at a fixed focus, we developed two methods, "Simfit" and "Focus Diversity" (W. F. Hoffmann, et. al. 1 ). Simfit finds the focus by obtaining the best match between observed images and families of simulated images at a range of focus settings. Focus Diversity utilizes the focal plane curvature to find the best fit of the varied image blur over the focal plane to a model defocus curve. Observations of a single star at many field locations in each of the four IRAC bands were analyzed before and during the refocus activity. The resulting refocus moves brought the focus close to the specified requirement of within 0.3 mm from the desired IRAC optimum focus. This is less than a "Diffraction Focus Unit" (λ×(f/ 2 )) of 0.52 mm at the SST focus at the shortest IRAC band (3.58 microns). The improvement in focus is apparent in both the appearance and the calculated noise-pixels of star images.