Hebrew relative clauses appear in two different forms, one containing a gap (trace) and the other with a resumptive pronoun (RP). RPs in Hebrew are often considered to be licensed only as a last resort strategy, on a par with intrusive pronouns in English (Shlonsky, 1992); nevertheless, such an approach fails to predict the availability of both traces and RPs within object relative clauses. This optionality has been analyzed as required for semantic/pragmatic purposes (Doron, 1982; Erteschik-Shir, 1992) or for general cognitive reasons (Ariel, 1999; Alexopoulou and Keller, 2002). In order to examine the functional role of RPs in relative clause processing, we conducted two acceptability judgment experiments, comparing the acceptability of RPs and traces in three environments: close to the relative head, in island environments, and embedded within an additional CP. Our results show that RPs are in fact more similar to traces than initially assumed. Specifically, both were rated lower in embedded clauses than in main clause relatives, and both were rated with low acceptability in island domains. Such resemblance between RPs and traces may stem from the fact that both are interpreted as bound variables, which makes them sensitive to the same set of processing constraints. By collecting quantitative ratings from multiple speakers on multiple examples, this empirical study thus complements theoretical analyses and identifies multiple factors contributing to the acceptability of Hebrew RPs and traces in object position.