This article addresses how good geoscientists are at assessing ultimate recoveries, production profiles, and value prior to development when compared to actual field behavior. Uncertainty studies conducted in the mid-1990s before production began on several United Kingdom Continental Shelf (UKCS) fields are re-examined on the basis of the actual outcomes, and practical guidelines based on the decade of new knowledge about the subsurface are proposed. Various subsurface uncertainties were systematically underestimated (e.g., compartmentalization) originally. This problem could be reduced in the future by considering a wider range of possible geological models and supporting this with modeling tools that enable the full breadth of uncertainty to be retained. The consistent use of information from analog fields may also have helped to appreciate the possible range of uncertainty more fully. I conclude that the essential difficulty of assessing uncertainty in field development is the integration of different exploration and production technologies along with the softer technology of human inference. Some uncertainties become more or less important during the natural life cycle of an oil field. For example, static volumetric issues (e.g., oil in place) generally decrease in importance whereas dynamic uncertainties (e.g., residual saturations, coning) increase. It should not therefore be a surprise that these dynamic issues become more relevant later in field life. This emphasizes the need during early production for robust surveillance plans that are properly executed to deliver the right data at the right time so that these uncertainties can be managed effectively.