Low statistical power challenges the reliability of animal research; yet, increasing sample sizes to the required level raises important ethical and practical issues. We present an alternative solution, RePAIR, which capitalizes on the observation that control groups in general are expected to be similar to each other. As shown in a simulation study, including information of previous control experiments in the statistical analysis using RePAIR reduced the required sample size by 49% or increased power up to 100%. We validated the potential of RePAIR in a unique dataset based on 7 independent experiments across the world, studying cognitive effects of early life adversity in mice. RePAIR comes with an open-source web-based tool (https://osf.io/wvs7m/) and can be widely used to largely improve quality of animal experimentation.