This paper reports on an ongoing research project called 'Prosodic Phrasing in Swedish', the object of which is to investigate prosodic phrasing and grouping in Swedish. Different methods exploited within the project are the analysis of speech production data, the use of text-to-speech synthesis and the use of speech recognition (prosodic parser). Production data from specially designed test material for Swedish has shown that tonal and temporal cues are combined to signal differences in phrasing. To test this analysis a perceptual experiment was conducted using the KTH rule synthesis where duration and F0 could be changed interactively. 12 listeners participated in the test with the task of identifying optimal positions for two distinct interpretations of an ambiguous test sentence as well as a line of ambiguity. The results of the experiment confirm our initial analysis and provide interesting individual variation indicating different perceptual strategies which may also be related to speaking habits. Longer texts are used in prosodic parsing experiments where the task is to identify prosodic phrases.