Not so long ago, uncertainty in economic evaluation was handled almost exclusively using simple one-way sensitivity analysis methods whereby individual parameters of an analysis were varied one-by-one over a range of (often ad-hoc) values. With the increasing use of the clinical trial as a vehicle for economic evaluation, there has been increasing interest in the use of statistical methods for handling uncertainty in patient-level data on both costs and effects. However, even for studies based on secondary data, the use of statistical methods for specifying parameter distributions in so-called probabilistic analyses is becoming increasingly common and is now the method recommended by a number of good-practice guides and regulatory agencies. The aim of this paper is to review these statistical approaches to the handling of uncertainty, both for primary studies where patient level cost and effect information is available and in secondary studies that typically use decision analytical methods to synthesize summary information from the published literature. The approach throughout the paper will be to emphasize the intuition behind the methods rather than the technical details. Although the methodology is generally applicable, examples will be drawn from both primary and secondary economic evaluations in the area of gastroenterology. The first example relates to a primary care trial of dyspepsia management which sought to estimate the cost-effectiveness of managing patients through early endoscopy compared to usual care. The second example is a secondary study of the management of gastro-oesophageal reflux disease symptoms which considered six potential strategies simultaneously.