This study investigated how prosodic features characterize Cantonese sarcasm and how these features help native listeners understand sarcastic meanings. 28 native Hong Kong Cantonese listeners listened to 50 sentences naturally produced by 6 native Cantonese speakers with a sarcastic attitude and with a sincere attitude. For each sentence the listeners rated whether they perceived the sentence as being produced with a very sincere (1) or very sarcastic (6) tone on a 6-point Likert scale. Acoustic analysis of the stimuli revealed that a slower speech rate, a lower mean F0, a lower mean amplitude, a narrower F0 range, a greater amplitude range, and a higher HNR value are all significant prosodic cues for Cantonese sarcasm although these cues may not be jointly present to deliver a sarcastic meaning. Listeners’ ratings indicated that Cantonese listeners were able to discriminate sarcasm from sincerity based on the prosodic features. The combination of the prosodic cues used by the speakers influenced how well the listeners could perceive sarcasm. The more prosodic cues were utilized in a sarcastic speech, the easier it would be for the listeners to understand the implied sarcastic meaning.