Morphological processing by French children was investigated in two experiments. The first showed that second and third graders read pseudowords such as "chat-ure" ("cat-ish") composed of an illegally combined real stem and real derivational suffix faster and more accurately than they read matched pseudowords composed of a pseudostem and a real derivational suffix (e.g., "chot-ure") or a pseudostem and a pseudosuffix (e.g., "chot-ore"). More, the chot-ure items were read faster and more accurately than the chot-ore items. These results suggest that beginning French readers are able to use morphological units (both stems and derivational suffixes) to decode new words. The second experiment compared the impact of display format on reading time. Suffixed words were presented in four segmentation formats: syllabic ("ma lade"), morphological ("mal ade"), morphological + 1 grapheme ("mala de"), or unsegmented ("malade"). For both groups of readers, the morphological + 1 condition generated the longest reading times but there was no difference between the other three conditions. It was concluded that syllables, morphemes, and whole word forms contribute to a similar extent to word reading for low-frequency words. Morphological processing may therefore be used early by French children to identify both new words and low-frequency words.