A challenge in functional magnetic resonance imaging (fMRI) is to develop imaging methods that are highly sensitive to microscopic field inhomogeneities [the blood oxygenation level-dependent (BOLD) effect] and minimally sensitivity to macroscopic fields. z-Shimming compensates for the through-plane dephasing that arises in gradient-echo images due to magnetic field inhomogeneities. To date, an analysis of the formation of composite images from multiple z-shim acquisitions has not been presented. This work compares three strategies for forming composite images, one of which is introduced for the first time, against the nominal image acquisition. True-versus false-positive rates of activation detection are considered, in addition to the time efficiency of the methods. It is shown that z-shimming can provide uniform spatial sensitivity, resulting in increased activation detectability, in many cases outperforming the nominal imaging approach. Time efficiency is shown to be dependent on field uniformity. Theory, computer simulations, and results from fMRI studies are used to demonstrate the performance of these methods.