This study aimed at investigating the guessing words-in-context strategies of 6 beginner and 6 upper-intermediate level EFL students at the College of Engineering and Architecture of Osmangazi University in Eskisehir, Turkiye. The data were collected through individual think-aloud protocols (TAPs) and retrospective sessions (RSs). In the TAPs, the participants were told to think aloud while they were guessing the five made-up test words. In the RSs, the participants later reported what helped them in their guessing. Both TAPs and RSs were tape-recorded. All protocols were transcribed. The transcriptions were examined to develop the taxonomy for the strategy types used; data-driven categorization was carried out. After that, frequencies of each strategy used were coded and calculated for each level of students. One of the major results revealed from the analyses of TAP and RS transcriptions showed that the beginning level participants used guessing words-in-context strategies more frequently than the upper-intermediate level participants although both groups frequently used the same strategies which were contextual clues and translation. The participants used morphological clues, phonological clues as well as the contextual richness of the passage. The results suggest that the use of guessing words-in-context strategies varied according to the clues that the test words or context offered rather than the proficiency level of the students. Therefore, guessing word-in-context strategies should not be avoided in beginner level instruction; these strategies can be taught and/or practiced without considering the proficiency level of the students. © Common Ground, G. Muge Kanatlar, Bena Gül Peker, All Rights Reserved, Permissions.