The purpose of this study was to investigate the roles of the putative "spatial filter" and "local sign" mechanisms in determining line vernier thresholds for a range of target separations, using stimulus contrast or visibility as a tool. In Expt 1, the effects of varying target contrast and exposure duration on vernier thresholds for lines separated by 90 min arc, where the reference line was fixated, were measured. Contrast thresholds for the nonfixated test Une were also measured, so that the role of its visibility in limiting vernier thresholds could be assessed. Vernier thresholds decreased almost proportionally with increasing contrast only until the visibility of the test line reached about 3 times the contrast detection threshold, regardless of exposure duration. At higher visibility levels, vernier thresholds were virtually independent of target contrast. In Expt 2, the effects of varying target contrast on vernier thresholds for a range of target separations (2-90 min arc) were measured using a 250 msec exposure duration. Vernier thresholds for abutting lines and for those separated by 2 min arc, decreased with target contrast until about 30 times the test line's contrast detection threshold. However for lines separated by 4 min arc or more, they were only weakly dependent on target contrast at much lower visibility levels. We propose that for very close separations vernier thresholds are limited by the contrast response properties of spatial filters. For separations of 4 min arc or more, thresholds appear to be limited by the positional uncertainty of the test line, which increases with eccentricity.