Uncertainty is intrinsic in most technical systems, including Cyber-Physical Systems (CPS). Therefore, handling uncertainty in a graceful manner during the real operation of CPS is critical. Since designing, developing, and testing modern and highly sophisticated CPS is an expanding field, a step to-wards dealing with uncertainty is to identify, define, and classify uncertainties at various levels of CPS. This will help develop a systematic and comprehen-sive understanding of uncertainty. To that end, we propose a conceptual model for uncertainty specifically designed for CPS. Since the study of uncertainty in CPS development and testing is still irrelatively unexplored, this conceptual model was derived in a large part by reviewing existing work on uncertainty in other fields, including philosophy, physics, statistics, and healthcare. The con-ceptual model is mapped to the three logical levels of CPS: Application, Infra-structure, and Integration. It is captured using UML class diagrams, including relevant OCL constraints. To validate the conceptual model, we identified, clas-sified, and specified uncertainties in two distinct industrial case studies.