The present study investigated whether and to what extent children with dyslexia utilize visual and phonetic strategies in character learning. A paired associate learning paradigm was used in two experiments to train children's pronunciation-orthography associations of novel words, with a recall task 1 week later for retention. Experiment 1 included 32 Mandarin-speaking fifth graders with dyslexia (dyslexia group) and 28 age-matched peers (comparison group) and manipulated the availability of an arbitrary bolded stroke in Chinese character (visual cue, available vs. unavailable) of eight low-frequency real characters. The dyslexia group demonstrated poorer character learning effects than the comparison group, whereas the similar interference effect of visual cues was found across groups. Sixty-six fifth-grade children participated in Experiment 2 (dyslexia, N = 34). The regularity of phonetic cues of 12 pseudo-characters was manipulated into regular, semiregular, irregular, providing full, partial, or no pronunciation cues. The dyslexia group demonstrated comparable learning outcomes of regular pseudo-characters, but poorer learning on semiregular and irregular pseudo-characters than the comparison group. Importantly, they utilize semiregular phonetic cues. In both experiments, the two groups did not differ on the retention of learning. Taken together, children with dyslexia perform poorer in the learning stage, but not in visual or phonetic strategies or the retention of learning. Like their peers, they do not use arbitrary visual cues but utilize phonetic cues, and thus compensate for poor learning of regular characters and alleviate that of semiregular characters.