The scientific literature is full of articles discussing poor reproducibility of findings from animal experiments as well as failures to translate results from preclinical animal studies to clinical trials in humans. Critics even go so far as to talk about a “reproducibility crisis” in the life sciences, a novel headword that increasingly finds its way into numerous high-impact journals. Viewed from a cynical perspective, Fett's law of the lab “Never replicate a successful experiment” has thus taken on a completely new meaning. So far, poor reproducibility and translational failures in animal experimentation have mostly been attributed to biased animal data, methodological pitfalls, current publication ethics and animal welfare constraints. More recently, the concept of standardization has also been identified as a potential source of these problems. By reducing within-experiment variation, rigorous standardization regimes limit the inference to the specific experimental conditions. In this way, however, individual phenotypic plasticity is largely neglected, resulting in statistically significant but possibly irrelevant findings that are not reproducible under slightly different conditions. By contrast, systematic heterogenization has been proposed as a concept to improve representativeness of study populations, contributing to improved external validity and hence improved reproducibility. While some first heterogenization studies are indeed very promising, it is still not clear how this approach can be transferred into practice in a logistically feasible and effective way. Thus, further research is needed to explore different heterogenization strategies as well as alternative routes toward better reproducibility in animal experimentation.