This dissertation is on the field of Psycholinguistics and focuses on the analysis of disfluencies in the speech of neurologically healthy older adults, using tasks from the Language Assessment Protocol called Bateria de Avaliação da Linguagem no Envelhecimento (BALE) (HÜBNER, et al., 2019). In this study, we aimed to investigate the most recurring types of disfluencies in the production of three types of narratives: a retelling (Lúcia), a funny story experienced by the participant (Funny Story), and a narrative based on a sequence of pictures (Dog Story). The study included 55 participants (43 female, 12 male), with a mean age of 70 years (SD = 7.3). We investigated the effect of the following sociodemographic variables on the production of disfluencies: age, gender, socioeconomic status, and education. We also examined the correlation between the proportion of disfluencies in each type of narrative and the score on the respective task, between the proportion of disfluencies and reading and writing habits, and the correlation between the proportion of disfluencies and the participants performance on cognitive tests included in the BALE. The results indicate that the proportion of disfluencies varies according to the cognitive demands of each type of narrative, with a higher proportion of disfluencies in the retelling task (Lúcia). The following types of disfluencies were analyzed: filled pauses, revisions, prolongations, repetitions, and interruptions. Differences were observed in the distribution of types of disfluency based on the grammatical nature of the affected item, with a higher incidence of filled pauses and revisions on lexical items and prolongations and repetitions on functional items. Regarding the examined sociodemographic factors, a significant result was found only for the education factor, with a higher occurrence of disfluencies among individuals with a higher level of education in the retelling task. Regarding the correlation studies, although there were no correlation effects between the proportion of disfluencies in each narrative and the score obtained in the respective task, it is worth noting that retelling was the task in which participants scored the lowest, which is in alignment with previous findings arguing that this type of narrative induces higher production of disfluencies. In the other correlation studies, a positive correlation was observed between the proportion of disfluencies produced in the funny story narrative and the naming task and between the proportion of disfluencies in the retelling task and the score obtained in the reading habits questionnaire. Taken together, these results suggest that disfluencies in the speech production of neurologically healthy older adults can be interpreted less as disturbances and more as important resources in the planning and linguistic structuring of utterances. Finally, implications of these findings are pointed out for a better understanding of the way elements such as narrative type, socioeconomic factors, reading and writing habits, and cognitive conditions/strategies interplay to shape the linguistic performance of older adults, bringing significant contributions to the field of psycholinguistics and beyond.