The purpose of this study was to determine differences in word recognition strategies used by good and poor readers. Twenty good readers and twenty poor readers were randomly selected from a fourth grade class and randomly assigned to levels of a five by five repeated-measures Latin Square design. All of the subjects were given two tasks to perform. The first task was to recognize tachistoscopically flashed words that were presented under five different treatment conditions. The other task was a modified cloze test in which prior context was given and only a minimal amount of visual information for the target word was supplied. The results indicated that good readers recognize words faster under all conditions and their recognition strategies were superior in that they were better able to generate words given minimal visual information. (WR)