We present results from an experiment studying how people mentally integrated partial configurations of objects shown across a sequence of displays with varying matches between frames of reference. Consistent with previous research on spatial updating, performance was better when the frame of reference in the final display aligned with the main display axes (up/down, left/right) than when it aligned with the diagonal axes. However, we also found that spatial updating was more efficient when the sequence of presentation of objects was consistent with the final frame of reference from which objects were integrated. Results suggested that spatial updating depended on the sequence of spatial operations required to integrate new spatial information into existing ones. Implications to theories of spatial updating in reasoning tasks are discussed.