In studies of masked translation priming with lexical decision tasks (LDT), an asymmetry is commonly observed: L1 translation equivalent primes can robustly activate their L2 targets, whereas priming effects tend to be smaller for L2 related primes (Wen & Van Heuven, 2017). The Revised Hierarchical Model (RHM; Kroll and Stewart, 2010) may explain the asymmetry by assuming differential access to the semantic information for L1 and L2 words (i.e., L2 lexico-semantic links are comparatively weaker, at least at lower proficiencies). This would hinder L2 words’ ability to access shared conceptual nodes, resulting in L2 translation primes being unable to activate their L1 equivalents in masked translation priming experiments. Enhanced L2-L1 priming effects are predicted with increased L2 development (e.g., higher L2 proficiency, more exposure to the L2). Under the Multilink model (Dijkstra et al., 2019), how often a word is encountered/used by each individual (i.e., subjective frequency) determines lexical processing speed. Comparatively lower processing speed prevents L2 primes to be successfully processed under masked priming conditions. Certain individual- and stimulus-level factors (e.g., L2 proficiency, L2 exposure/use, or L2 word frequency) might proxy subjective frequency and predict modulations in translation priming effects. To examine these claims, we tested 60 late sequential L1 Spanish-L2 English bilinguals living in an L2-environment in a masked translation priming LDT. A 500 ms mask was followed by a 60 ms display of the prime, immediately followed by the target. Two individual-level factors (L2 proficiency and amount of L2 exposure/use) and a stimulus-level predictor (word frequency) were examined and treated as continuous variables in linear mixed-effects models (Baayen, 2008). This design attempted to weigh the role that these factors have in translation priming effects, both individually and in interaction with each other, contributing to the dearth of studies on the priming asymmetry treating these variables continuously. Our results do not replicate the asymmetry (L1-L2: 38 ms priming; L2-L1: 39 ms priming; difference not significant). Response times in the critical condition were significantly faster than in the unrelated condition in both translation directions, suggesting that the related primes in both languages were processed efficiently and activated their targets. Crucially, the L2-L1 priming effects were modulated by a significant effect of Language exposure/use, those participants with larger amounts of active L2 use showed larger L2-L1 priming effects. L2 proficiency and word frequency only showed an effect in complex interactions. Our main finding is unable to tease apart the predictions of the RHM and Multilink. The present findings call for more research on how subjective frequency might be efficiently proxied through both stimulus- and individual-level factors.