When a target is surrounded by a four-dot mask pattern that persists after the target disappears, target identification is worse than when the mask terminates with the target. This masking effect has been attributed to Object Substitution Masking (OSM). Previewing the four-dot mask attenuated OSM. This study investigated specific situations in which mask preview was (or was not) effective in attenuating masking. In Experiment 1, the interstimulus interval (ISI) between previewed mask offset and target presentation was manipulated. The basic preview effect was replicated; neither ISI nor preview duration influenced target discrimination performance. In Experiment 2, mask configurations were manipulated. When the configuration of the mask at preview matched the configuration at target presentation, the preview effect was replicated. New evidence of ineffective mask preview was found: when the two configurations did not match, performance declined. Yet, when the ISI between previewed mask offset and target presentation was removed, such that the mask underwent apparent motion, preview was effective despite the configuration mismatch. The object-token interpretation provides an excellent account to this data. This study enabled a new understanding of when preview exactly attenuates OSM.