The present study examined the processing and performance of examinees in reading comprehension (RC) tests, when they read the whole text "prior" to its questions ("text-first" strategy [TFs]) compared to reading the text "while" and "for" answering the questions ("questions-first" strategy [QFs]). Previous studies have been inconclusive regarding the benefit of TFs, and no studies to date have indicated a benefit for QFs, although test-taking guidebooks and test-preparation courses and websites regularly recommend the latter strategy. This study examined whether and how test time constraints and examinees' domain knowledge modulate the effect of TFs and QFs on examinees' performance accuracy, performance time, and text search efficiency in RC tests, when they answer high- versus low-level questions. University students with and without domain knowledge were asked to read two expository texts and answer open-ended questions under TFs and QFs settings, with or without time limitation. Findings demonstrated higher performance accuracy, shorter performance time, and lower search efficiency for QFs as compared to TFs setting, particularly when the task was timed. Domain knowledge enhanced performance accuracy regardless of strategy and time constraints, although performance time and search efficiency were similar for high- and low-knowledge groups. These findings suggest that university students should apply QFs in timed tests with open-ended questions, and demonstrate the manner in which both strategies can be implemented effectively under different conditions. The theoretical implications of the present findings for goal-oriented models and RC tests' validity are further elaborated in the Discussion section.