This study brings empirical data from Hebrew to support Selkirk's claim that prosodic phrasing is the result of an interaction of syntactic and phonological constraints. Employing a recently devised protocol, a production experiment elicited contextually disambiguated utterances of the relative clause attachment ambiguity construction. Satisfying the syntactic alignment constraint (Align R XP) and the phonological length constraints (BinMin and BinMax) would result in different prosodic phrasings for this syntactically ambiguous construction, permitting a test of the ranking of these constraints. Two types of data analysis suggest that in Hebrew the syntactic and phonological constraints are strictly ranked: In a context where the two types of constraints are in conflict, the alignment constraint is dominant; when the alignment constraint is vacuously satisfied, effects of the lower-ranked length constraints come into play. Together, these experimental findings illustrate the descriptive value of the constraint ranking hypothesis for prosodic phrasing.