The first goal of this study was to examine how individual differences in initial L2 proficiency help explain L2 grammar development in oral production during short-term immersion abroad. The second goal of the study was methodological, and evaluated challenges that can result from operationalizing learners' initial L2 proficiency as pretest performance on outcome measures (as opposed to independent proficiency measures) in analyses of L2 change. L2 Spanish learners participating in summer study abroad completed an elicited imitation task and two oral production tasks. Production data were analyzed for changes in relevant grammatical complexity and accuracy dimensions. Results indicate that learners with higher initial L2 proficiency experience greater L2 grammar advancement from short-term immersion, and that pretest performance can be an unreliable operational estimate of initial proficiency when analyzing L2 gains. We discuss findings following cognitive accounts of SLA, and highlight methodological implications for further research in immersion contexts and beyond.