This study first aimed to investigate disfluency clusters in typical and atypical Finnish adult speakers. Secondly, it aimed to observe possible fluency strategies in speakers representing different fluency levels. In addition to individual disfluency types, we examined different characteristics of disfluency clusters produced by 23 speakers in a fluency continuum. Three adult speaker groups participated in this study: typical speakers with high disfluency frequencies (GA), typical and atypical speakers with very high disfluency frequencies (GB) and atypical speakers with the highest disfluency frequencies (GC). Data were based on a narrative speech task, and disfluency clusters were analysed with both traditional methods and alternative methods. Two statistically significant differences between the speaker groups were found: 1) the length of the clusters was highest in GC compared to other groups, and 2) speakers in GC formulated their utterances more than other groups. Other results, although nonsignificant, were that 3) speakers in GA revised utterances more often than interrupted them compared to GB and GC speakers, and 4) clusters using repetitive words and phrases to maintain fluency were found in GA and GB only. In this study, different fluency levels revealed different strategies in both the production of single disfluencies and in disfluency clusters. It seems that more fluent speakers formulate their messages differently than less fluent speakers, and repetitions can be used to maintain fluency and possibly prevent difficult clusters, as noted with more fluent speakers.