This study investigates the hypothesis that variations in symbol sign legibility distance can be accounted for on the basis of a sign's dependence upon high spatial frequency contours to convey critical information. Using digital image processing techniques, highway signs were blurred to remove all high spatial frequency information. A blur recognition threshold was established for each experimental sign by sequentially “deblurring” it until the observer could report the critical details defining its recognition criteria. Correlational analyses were then conducted to determine if legibility distance (collected in a previous study) could be predicted from the blur recognition threshold data. A significant correlation was observed between blur recognition threshold and sign legibility distance (r = −0.734, N=12, p < 0.001). That is, symbol signs with high levels of “blur tolerance” could be recognized at significantly greater viewing distances. These results support the application of new computer-assisted “recursive-blur” design techniques to optimize the effectiveness of symbol highway signs and related visual stimuli (see Schieber, Kline and Dewar, 1994).