Imitation tasks are used in intonation research to identify properties that are perceptually salient and encoded for subsequent production. The current study examines whether and how imitation of synthetic versus naturally produced F0 contours may differ. We compared F0 contours in American English from two imitation experiments where participants heard sentences with the same phrase-final intonation and reproduced the heard pattern on a novel sentence. In one experiment, F0 patterns of stimuli were controlled via pitch resynthesis using straight-line approximations of (phonological) tonal targets; the other used natural productions of the same tunes. F0 trajectories were examined to identify which F0 properties of the stimuli were preserved or lost as a function of the type of stimulus. Imitations of natural vs. resynthesized stimuli were compared using time-series k-means clustering analysis, GAMM modeling, and RMSD as a measure of F0 difference between imitation and stimulus. We observe striking similarity in imitations of natural and resynthesized stimuli based on clustering solutions, with small, localized differences in GAMMs for only two out of eight tunes tested. RMSD results show closer imitation with resynthesized stimuli, suggesting greater attention to fine phonetic detail of F0 patterning when other cues to intonational contrasts are held constant.