This study investigates the phonetic realizations of Korean vowels /ㅔ/ and /ㅐ/ produced by Chinese and English learners of Korean at beginner and advanced levels of proficiency. Further, it tests the extent to which acoustic similarity between L1 and L2 sounds is crucial in predicting easy/difficult perception and production of L2 sounds. Although Korean vowels /ㅔ/ and /ㅐ/ used to be two separate phonemes in Korean, Korean speakers no longer distinguish them due to the merger of these two vowels. The results from this production study reveal two findings. Firstly, the L1 influence on L2 production is partially supported for learners of Korean in that English learners of Korean do not differentiate Korean [에] and [애] in durational and spectral cues in their production. Secondly, the advanced Chinese learners exhibited hypercorrection in that they produce the Korean vowel /ㅔ/ very distinctively from /ㅐ/. These findings suggest that acoustic similarity alone cannot predict the likelihood of easy/difficult production of L2 sounds contra Flege’s (1995) Speech Learning Model (SLM) and Escudero et al.’s (2004) Second Language Linguistic Perception Model. They indicate that teaching method also plays a crucial role in second language acquisition.