The reproducibility of the results from preclinical research rests on many factors, including the selection of appropriate experimental designs for the individual experiments that constitute the investigation. The design of each of these experiments depends on their purpose within the entire investigation and the information to be gained from conducting them. Here, we explain and justify a three-stage strategy comprising a series of different types of experiment, each with a different purpose and design: a pilot study, a hypothesis-generating experiment and a final hypothesis-confirming experiment. Compliance with this three-stage strategy, over the course of an entire investigation, will not only strengthen its reproducibility but, importantly, can save time and other resources, including the total number of animals used.