The visual system filters spatial pattern through a range of narrowly tuned spatial mechanisms, but the rules by which the outputs of these mechanisms are combined across time to extract precise geometrical information are not yet clear. One hypothesis is that spatial analysis shifts over time from coarse to finer spatial scales, in order to extract fine spatial information. An alternative hypothesis is that thresholds are determined by the signal-to-noise ratio within an optimal spatial scale. In this study, we measured vernier acuity across exposure duration for equally visible long lines and short lines and found no improvement in spatial precision with time. Using a simultaneous spatial-masking paradigm, we determined the active spatial scales at 100 and 1000 ms. The results show no significant changes in spatial scale, or in the size-range of active scales, for the two exposure durations. Furthermore, whereas vernier thresholds vary markedly with line contrast, we find only modest shifts in spatial scale. Taken together, our results suggest that for vernier acuity, spatial scale is selected very early, and that vernier thresholds are predominantly limited by signal strength within that spatial scale.