The time-course of morphological processing during spoken word recognition was investigated using event-related brain potentials (ERPs) in an auditory lexical decision task. We compared three different types of French words: truly suffixed (e.g., pochette ‘little pocket’ = poche ‘pocket’ + diminutive suffix -ette), pseudo-suffixed (e.g., mouette ‘seagull’ = mou ‘soft’ + pseudo-suffix -ette) and non-suffixed target words (e.g., fortune ‘fortune’ = fort ‘strong’ + non-suffix -une). Suffixed (e.g., mouesse = mou + suffix -esse) and non-suffixed nonwords (e.g., mouipe = mou + non-suffix -ipe) were also tested. The behavioural results showed that participants responded more slowly to non-suffixed words than to truly suffixed and pseudo-suffixed words, but there was no difference between these two suffixed conditions. Moreover, participants made more errors rejecting pseudo-suffixed nonwords than non-suffixed nonwords. In the ERP analyses, T0 was shifted to the end of the embedded stem or pseudo-stem. The ERP results revealed enhanced N400 amplitudes for non-suffixed words compared to truly suffixed and pseudo-suffixed words. Again, there was no difference between the truly and pseudo-suffixed conditions. In addition, we found an increased N400 amplitude for both pseudo-suffixed and non-suffixed nonwords than for words. The latency of the onset of this N400 effect varied between the three experimental conditions: the word-nonword difference occurred earliest in the truly suffixed condition, slightly later in the pseudo-suffixed condition and latest in the non-suffixed condition. Both behavioural and EEG data jointly suggest that spoken words with a genuine morphological structure and words with a pseudo-morphological structure are decomposed into morphemic sub-units. Moreover, the earlier appearance of the N400 effects in the truly suffixed condition indicates that morphological information is more readily available in words with a semantically transparent morphological structure.