To slow the spread of COVID-19, many universities have had to move to online teaching, which entails changing exams from in-person to online. Online exams can facilitate cheating when there is no direct proctoring. To provide some form of control in unproctored exams, Cluskey et al. (2011) suggested having substantial time pressure; yet there are no clear recommendations on how to implement it. We conducted an experiment comparing three implementations of time pressure (individual allocation of time, equal time for each task, and time-based on task length) in a low-stakes educational sciences exam. Students differed in their answering strategies and objective performance but not in their evaluations of the different conditions. We suggest having equal time for each task to maximize students' performance with minimal opportunities to cheat.