We report the results of whether syllables are frequency-modulated prelexical units in dyslexic children. Twenty-two French dyslexic children were compared to 44 chronological age-matched and reading-level-matched controls. A syllable compatibility procedure was combined with a visual syllable detection task (Exp. 1) and a visual masked priming paradigm in a lexical decision task (Exp. 2). Dyslexic children exhibited robust frequency-modulated prelexical syllable effects; high-frequency syllables elicited a syllable compatibility effect in both experiments, while low-frequency syllables favored either a CV target length or a CVC prime length effect. The frequency-modulated syllable effects were constant across both experiments following an expected developmental course, especially in a highly feasible task (Exp. 1). However, performance was drastically low in a highly demanding task (Exp. 2), suggesting impaired phonological procedures. We propose that dyslexic children do not have obvious impaired phonological representations but rather delayed or compensated phonological representations with impaired phonological procedures.