Many studies (Bern 1995; Herron 1994) suggest the help of pre-listening questions with comprehension. Sherman (1997) even explores the best procedure to present such questions. Yet, her study does not take into account participants’ proficiency levels and the nature of questions. The present study, therefore, first aims to investigate the effects of three different test procedures (Questions Before [QB], Questions After [QA], Sandwich [S]) on EFL learners’ listening comprehension. Then, it also explores procedures’ interaction with learners’ proficiency levels and question types. In this study, 171 university students were recruited and classified into high, intermediate, and low levels of proficiency based on their scores of a TOEFL listening test. They then took another listening test which encompassed three texts, with each text containing three question types: global, local and inferential. The primary distinction of the three procedures lies in the timing of question provision. As its name suggests, Questions Before (QB) distributed questions for previewing prior to the two hearings. By contrast, Questions After (QA) postponed the questions until the last whereas Sandwich (S) offered the questions for the preview between two times of listening. In the study, after the test a questionnaire was also employed to elicit participants’ attitude toward test procedures. The results indicated an effect of test procedures on listening comprehension, with S outperforming QB, which in turn surpassed QA. Yet, there was not a significant relationship between test procedures and proficiency or question types. Global questions roughly displayed higher scores in both QB and S. In contrast, inferential and local questions exhibited lower scores in S. The questionnaire responses showed participants’ preference for S, then QB and finally QA. This ranking approximately consented to the test results. Based on the findings, some pedagogical suggestions were made.