Recently the observation of low-frequency microtremors using passive seismic surface observations was proposed as a direct hydrocarbon indicator. Applying reverse time modeling to such data indicates that observed spectral anomalies from the microtremor data originate from the reservoir area. Here we use a modified diffraction stack to locate synthetic microtremor events. The diffraction stack provided good results for the localization of acoustic emissions (i.e., transient high-frequency signals). Applying the stacking technique to low-frequency microtremor data indicates limitations of this method. For strongly heterogeneous media and low-frequency data we have interfering events due to triplications. The moveout of these events is different from the moveout of first arrival traveltimes used in the diffraction stack. This leads to a focusing of the energy at a wrong depth position. The source images using the stacking approach have an improved S/N ratio for shallow levels when compared to the localization o