Among many cognitive deficits, patients with Alzheimer's disease (AD) show reduced working memory capacity (Baddeley et al., 1991), simplified syntactic production (Lyons et al. 1994), and reduced comprehension for complex sentences (Mcdonald et al., 2001). Although the relationship between working memory capacity and language comprehension is clearly established (Almor et al., 2001; Just & Carpenter, 1992), the relationship between working memory and language production is much less ascertained (Alamargot and Chanquoy, 2002). However Kemper & Summer (2001) found that syntactic complexity correlates positively with backward digit span. 49 participants diagnosed with Alzheimer's disease and 31 healthy controls matched for age, level of education and socioeconomic status participated in the study. Spoken production was elicited using the Cookie Theft Picture Description Task (BDAE) (Goodglass et Kaplan, 1976). Participants' speech was transcribed in CHAT format (MacWhinney, 2001) and several complexity measures were examined: MLU (Brown, 1973), MCU, (Cheung et Kemper, 1992), proportion of simple, coordinate and subordinate clauses. In addition, patients' comprehension was tested with the Token test (De Renzi & Vignolo, 1982) and short term and working memory spans were assessed using forward and backward digit spans (Wechsler 1987). Results show that AD patients produced significantly shorter MLU and MCU, as well as significantly fewer coordinate and subordinate clauses. AD patients also scored significantly lower on the Token test. We found no difference between patients and controls for the forward digit span task, but the patients showed significantly lower backward digit spans. Finally, backward digit spans correlated significantly with comprehension scores while in production only MCU showed a significant correlation. These results suggest that syntactic complexity in production does not rely solely on working memory capacity.