A letter counting task was presented on a Macintosh computer screen using two different versions of 24 point Times Roman print. One version, called “grainy,” had 12 matrix units per font height and the other, called “smooth,” had 24. A mixed group of low vision subjects and a normally sighted group had speed and accuracy measured as they performed the test task from two different viewing distances. For both population groups, the remote test distances were arranged individually so that the letters subtended an angular size scarcely larger than threshold. The close distances were fairly representative of practical working distances. It was shown that, for both groups of subjects, smooth letters allowed faster performance for the closer working distances only. Smoothing the letters helped accuracy of performance at the far distances, and at the close distances, smoothing improved the accuracy for the low vision group but not for the normals. The implications are discussed.