INTRODUCTION The adoption and release of genetically modified (GM) crops is the focus of a contentious debate. On the one hand, this new technology has been adopted rapidly by many producers. It is estimated that 40.5 million hectares were planted worldwide in 1999 (The Economist 1999), an increase of 13 million hectares from 1998 and four times the amount planted in 1997 (James 1998). The popularity of GM crops is owed to genetic modifications that have added traits such as herbicide tolerance and pest resistance that lower average input costs (ERS 1999). On the other hand, despite broad acceptance in some farming circles, and the potential for numerous environmental, health and other benefits, the release of GM crops is the subject of international scrutiny and controversy regarding their safety (see in this issue, Perdikis and Kerr; Hobbs and Plunkett). Of particular concern in both the popular press (e.g., Bonham 1999) and some scientific articles (Butler et af. 1999) is the potential for the widespread release of GM crops to result in ' environmental externalities, including irreversible environmental damage. In part, this view is fostered by concerns that GM crops are being released too quickly and without adequate knowledge oftheir long term environmental impacts (Mander and Goldsmith 1996; Butler et al. 1999) or, in some countries, without procedures to deal with environmental problems that may occur (Hruska and Lara Pavon 1997). Currently the decision to release a GM crop within Canada or the United States is based on an assessment of the risks of environmental, health and other potential damages. Although considerable research is being undertaken to understand these risks, opinions differ regarding the probability of environmental damage. In this paper, a simple theoretical framework is used to demonstrate that the current decision framework could be improved by considering additional factors such as economic benefits and potential costs. In some cases, this additional information could lead to decisions that are the polar opposite ofthose suggested under the existing risk assessments. It is demonstrated that even when the probability of an adverse event occurring is high (low), there may be net benefits (net costs) from releasing the GM crop. Releases of GM crops are complicated by the potential for irreversible environmental damage that can change the long term profitability of production decisions. The frameworks presented clearly demonstrate that economic analysis can be used to improve current decisions to release GM crops.