The present study investigated the efficacy of segmental/suprasegmental vs. holistic pronunciation instruction in the development of listening comprehension skills by EFL learners, using a pre-test post-test design. Six groups of 20 intermediate EFL learners at a university in Iran took part in the study, all groups receiving the same amount of instruction (10 hours over 5 weeks). The control group listened to/viewed authentic audio recordings and movies in English, discussed their contents, and completed a variety of listening comprehension tasks but received no pronunciation instruction. Four experimental groups completed similar activities but during one third of the teaching time (20 minutes per class), received an explanation of segmental or suprasegmental features followed by production-focused or perception-focused practice. The final experimental group received holistic pronunciation instruction with mixed perception/production-focused practice for 20 minutes during each hour-long class. Versions of Longman's TOEFL English proficiency test (paper-based) were used to assess listening comprehension at pre-test, immediate post-test and delayed post-test. The findings revealed that the holistic pronunciation instruction enhanced the listening comprehension skills of Iranian EFL learners more than separate segmental or suprasegmental training, with either perception or production-focused practice.