Listener-based judgements of fluency play an important role in second language (L2) communication contexts and in L2 assessment. Accordingly, our meta-analysis examined the relationship between different aspects of utterance fluency and listener-based judgements of perceived fluency by analyzing primary studies reporting correlation coefficients between objective measures of temporal features and subjective ratings of fluency. We analyzed 263 effect sizes from 22 studies (N = 335-746) to calculate the mean effect sizes of the links between utterance and perceived fluency. We also investigated the moderator effects of 11 methodological factors--such as speech stimuli, listeners' background, rating procedure, and computation of utterance fluency measures--on the relationship between utterance and perceived fluency. Perceived fluency was strongly associated with speed and pause frequency, r = |0.59-0.62|; moderately with pause duration, r = |0.46|; and weakly with repair fluency, r = |0.20|; while composite measures showed the strongest effect sizes, r = |0.72-0.76|. Moderator analyses revealed that the utterance-perceived fluency link is influenced by methodological variables related to how speech samples are prepared for listeners' judgements and how listeners' attention is directed in evaluations of fluency. These findings suggest future directions for L2 fluency research and implications for language assessment.