A popular paradigm for studying prominence is Rapid Prosody Transcription (RPT), in which linguistically untrained participants listen to utterances and mark on a transcript words they perceive as prominent. RPT responses can be sensitive to the type of instructions used, such as whether participants are asked to select only the most prominent word or all the words they deem prominent. Here, we compare results from two RPT studies with the same Greek materials but using the two above-mentioned instruction types. Inter-rater agreement scores were similar across the two studies and yielded comparable overall results (e.g., words in focus were more likely to be selected in both). However, the relevance of certain criteria varied depending on task: in the multi-word task, duration, amplitude, and F0max predicted prominence, while in the single-word task F0max was not a predictor. These differences suggest that the tasks investigated here are not interchangeable, despite similarities. Most importantly, they can each lead to different interpretations of what constitutes prominence, suggesting that researchers need to be cautious when using them.