Recently, increased interletter spacing (LS) has been studied as a way to enhance reading fluency. It is suggested that increased LS improves reading performance, especially in poor readers. Theoretically, these findings are well substantiated as a result of diminished crowding effects. Empirically, however, findings on LS are inconclusive. In two experiments, we examined whether effects of increased LS are specific to children with dyslexia and whether increased LS affects word or sentence processing. In the first experiment, 30 children with dyslexia and 30 controls (mean age=9years 11months) read sentences in standard and increased LS conditions. In the second experiment, these sentences were read by an unselected sample of 189 readers (mean age=9years 3months) in either a sentence or word-by-word reading condition. The first experiment showed that increased LS affected children with dyslexia and controls in similar ways. Participants made fewer errors in the increased LS condition than in the standard LS condition. Reading rates were not affected. There were no indications that the effect of LS was related to reading ability, not even for a subgroup of readers. Findings of the second experiment were similar. Increased LS resulted in fewer errors, not faster reading rates. This was found only when complete sentences were presented, not when sentences were read word by word. Three main conclusions can be drawn. First, increased LS appears to affect reading accuracy only. Second, the findings do not support claims that increased LS specifically affects poor readers. And third, the effect of LS seems to occur at the interword level. Theoretical and practical implications of these findings are discussed.