Lardiere’s (2005, 2008, 2009) Feature Reassembly Hypothesis proposes that L2 acquisition involves reconfiguring the sets of lexical features that occur in the native language into feature bundles appropriate to the L2. This paper applies the Feature Reassembly Hypothesis to findings from recent research into the L2 acquisition of existential quantifiers. It firstly provides a feature-based, crosslinguistic account of polarity item any in English, and its equivalents — wh-existentials — in Chinese, Korean and Japanese. It then tests predictions built on the Feature Reassembly Hypothesis, about how learners map target existential quantifiers in the L2 input onto feature sets from their L1, and how they then reassemble these feature sets to better match the target. The findings, which are largely compatible with the predictions, show that research that focuses on the specific processes of first mapping and then feature reassembly promises to lead to a more explanatory account of development in L2 acquisition.