A small increase in inter-letter spacing improves legibility of texts, especially for dyslexics. Reduced crowding is one possible source of this improvement. Here we aimed to determine whether this improvement might also be linked to greater inter-letter spacing leading to more precise saccade computation. In two experiments, we recorded eye movements while dyslexic children, reading-level controls-RL, and chronological-age controls-CA, performed an oculomotor lateralized bisection task on words and strings of hashes displayed in a non-proportional font. In Experiment1, stimuli of 5 or 6 characters in length were presented in a spaced or non-spaced condition. In the non-spaced condition, dyslexics and RL performed similarly, both showed a weaker left/right asymmetry than CA. Conversely, in the spaced condition, RL were more affected by the extra spacing than dyslexics. We ran a second experiment on CA to disentangle the role of inter-letter spacing and spatial width. Stimuli were made the same length in the spaced and non-spaced condition by reducing the number of characters in the spaced condition (5-spaced vs. 9-non-spaced characters). We observed longer first saccades for 9-non-spaced characters presented in the left visual field, suggesting an important role for number of characters in guiding saccadic computation when controlling for spatial width.