A call to action to address critical flaws and bias in laboratory animal experiments and preclinical research.

During the design of hypothesis-driven, comparative laboratory animal experiments, investigators must control for cage effects, ensure full blinding and full randomization while adhering to established experimental designs, notably variations of the Completely Randomized Design and the Randomized Block Designs. Failure to meet these criteria introduces partial or complete confounding by multiple known and unknown variables, resulting in biased outcome measures and rendering valid statistical analysis impossible. Our analysis of a stratified, random sample of comparative laboratory animal experiments conducted in North America and Europe and published in 2022, shows that as few as 0-2.5% utilized valid, unbiased experimental designs. The failure of investigators to adopt valid, unbiased study designs undermines scientific rigour, squanders resources and animal lives, and impedes the reliable translation of preclinical research findings to human and veterinary medicine. We propose practical, achievable solutions focused on enhancing the rigour and validity of study designs. This includes developing a specialized group of scientists with expertise in the design of laboratory animal experiments and data analysis, to ensure future studies are conducted with the highest scientific standards.