The goal of this paper is to observe revision during handwritten text production of French students with and without dyslexia. Subjects with typical language development automate spelling during childhood and adolescence, progressively with experience, this enables them—according to capacity theory applied to written text production (McCutchen, Educational Psychology Review, 8, 1996, 299)—to allocate more cognitive resources to higher‐level processes (Bereiter & Scardamalia, The psychology of written composition. Hillsdale: Lawrence Erlbaum, 1987). A lack of automation in spelling could result in poor compositional performance (Fayol & Miret, Psychologie Française, 50, 2005, 391). Moreover, Morken and Helland (Dyslexia, 19, 2013, 131) have shown that young children with dyslexia, engaged in a sentence dictation task, revise their text as much as control children, however their revisions are of lower quality. If students with dyslexia have not totally automated spelling (Mazur‐Palandre, Développements (Revue Interdisciplinaire du développement Cognitif Normal et Pathologique), 18, 2016, 177), what is the impact on higher‐level processes, such as revision? Our goal is to examine if students with dyslexia and control students proofread their texts in the same way. Results show that they display some similarity but students with dyslexia seem to have a deficit in the error detection mechanism (Horowitz & Breznitz, International Journal of Disability, Development and Education, 58, 2011, 33) and revisions are less efficient. We discuss these results by considering previous studies we conducted on spelling, speech and neuropsychological assessments of our dyslexic participants.