How does science manage uncertainty? It is an inherent and necessary feature of knowledge production—especially when science informs policy. Drawing on examples from climate science, medical research, and public health, we can distinguish between different sources of uncertainty, including measurement error, model assumptions, statistical inference, and epistemic biases. We revisit the philosophical foundations of scientific uncertainty, from Popperian falsifiability to Bayesian inference, and emphasises how scientific norms—controls, replication, peer review—aim to reduce error while remaining open to revision. The reproducibility crisis is introduced, which highlights how misuse or misunderstanding of statistics (e.g. p-hacking, conflating correlation with causation) can undermine trust in scientific findings. Increasing reliance on complex models and machine learning introduces new challenges in transparency and validation, especially when these tools are used to guide policy under conditions of high stakes and incomplete information. The precautionary principle and the concept of post-normal science are introduced as frameworks for navigating risk in policy decisions. Ultimately, the chapter shows that while science embraces uncertainty as part of discovery, policy prefers certainty—creating persistent tensions at the interface of science and power.