This study set out to explore the comprehension of formulaic sequences, -particularly body-part idioms, by Saudi EFL learners. The study is essentially empirical and is situated within a cognitive linguistics-inspired framework. This approach goes hand in hand with construction grammar (cf. Fillmore et al. 1988). This approach proved to be plausible as it treats idioms as a central part of the language, in contrast to traditional approaches, e.g. generative grammar (e.g. Chomsky, 1980) or structural linguistics (e.g. Hockett, 1958) which marginalise idioms as language oddities or part of the 'periphery' but not the 'core'. A total of 91 Arabic-speaking female participants, majoring in Languages and Translation at King Saud University (KSU), Saudi Arabia were recruited for this study. The performance of the participants on idioms - moderated by language proficiency - was assessed in relation to four variables: similarity of idioms to L1, their level of compositionality, their level of transparency, and their frequency in the BNC. The language proficiency was assesed using the Oxford Placement Test (OPT). A test of idioms in a Multiple-Choice Question (MCQ) format was devised, piloted and validated. The test aimed to assess the receptive knowledge of the learners on idioms related to body parts. In addition to this, the test aimed to investigate the factors that affect the performance of the participants on the test of idioms at the comprehension level. The initial criterion of selection for the items was based on similarity to the first language (LI) of the participants (Arabic). A total of 60 items were included, 30 of which were similar to Arabic, and 30 of which were dissimilar. The variables pertaining to the characterizing features of idioms were carefully operationalized. In relation to similarity to L I, two measures were undertaken. First, idioms were categorized based on: (1) the similarity/dissimilarity of the idiom at the linguistic level; (2) the conceptual level that motivates the idiom. Second, native speakers of Arabic were required to judge how similar the idioms are to Arabic on a 5-point Likert scale. The same procedure of operationalization was followed to measure compositionality and transparency, following specific definitions and criteria. As for the frequency of idioms, this was checked from the BNC, following certain formulae so as to capture all possible instances. The data were analyzed following a mixed method design - combining quantitative and qualitative paradigms. The quantitative analyses included running a test for correlation between language proficiency and the overall scores of the learners on the test of idioms, as well as on the correlation between the four variables and the overall scores of the participants on the test of idioms. The qualitative analysis involved the use of think-aloud (TA) protocols, which aimed to tap into some additional factors, explore the strategies employed to understand idioms and also triangulate the results. The findings reveal that language proficiency and the total scores on the test of idioms correlate significantly. Moreover, judgements of the similarity to Ll and transparency judgements by Arabic speakers also correlate significantly with the overall scores on the test of idioms. However, compositionality judgements and frequency hits did not yield any statistically significant correlation. Further, the TA protocols corroborate the quantitative analysis and reveal that language proficiency affects the performance of the participants, as do transparency and similarity to Ll. Abundant cross-linguistic influence (CLl) was noted on the verbalization of 10 participants, who also displayed some discrepancies across different proficiency levels. The findings also show some interesting findings related to ,properties of body-part lexis, such as imageability, concreteness and translatability, as well as familiarity