BACKGROUND: Among many techniques available for pharmacoeconomic analysis (PEA), some, such as decision analysis (DA), have been formally defined, whereas others are still in the early stages of development and are less well defined. Used for quite some time in the business sector, DA has been adopted for PEA for its apparent straightforward nature and ease of use. METHODS: To evaluate the utility of DA techniques for PEA, an analysis of treatment for gastroesophageal reflux disease (GERD) will be presented as an example and the basic tenets of DA examined. RESULTS: Although the basic decision tree for the GERD model appeared straightforward, it became clear upon implementation just how difficult it is to fit disease management and progression into such a model. Instead of the standard two alternative-two outcome textbook examples, there were six initial choices, three potential early outcomes and countless subsequent resulting "next steps". As with most PEA, it was also necessary to explicitly account for time as many model elements depended on time. CONCLUSIONS: Among problems with DA are: time is not explicitly accounted for, the use of dual values (cost, effectiveness) was not intended, and cost and effectiveness values must be specified at a terminal node though they are actually cumulative. Thus, DA is clear only for simple models — more often it is overly complex, computationally inefficient and very difficult to validate. Techniques better suited to PEA will be discussed. Any person who uses, or is interested in using, PEA in his or her work will benefit from attending this session.