Butler (1980) compared errors representing intrusions and mislocalizations on 3x3 letter displays under pattern-mask versus no-mask conditions and found that pattern masking increased character mislocalization errors (naming a character in the display but not in the target position as being the target) over intrusion errors (naming a character not in the display as being the target). The possibility that pattern masking may increase feature mislocalizations was examined in this study. A small rectangle was superimposed on a field of vertical colored strips and the exposure of each rectangle was terminated by either a pattern mask or a blank exposure field superimposed upon the continuously-present field of strips. Each of 12 subjects received both postexposure conditions in a random order over trials. Significantly more mislocalizations occurred with pattern masking; mislocalizations toward fixation predominated. The tendency for mislocalizations to proceed inward increased significantly with eccentricity, but significantly more so with pattern masking. This study confirmed Butler's observations in a pure localization experiment, and interpreted in consonance with feature perturbation theory. (ABL)