A procedure is presented for developing environmental risk hypotheses associated with thedeployment of forest trees genetically modified to have altered wood properties and forselecting non-target species to test these hypotheses. Altered-lignin Pinusradiata trees intended for use in New Zealand are used as a hypothetical casestudy to illustrate our approach. Firstly, environmental management goals (such as woodproduction, flood control or preservation of biodiversity) were identified and linked tothe forest attributes they require. Necessary conditions for each attribute were listedand appropriate assessment endpoints for them developed. For example, biological controlof pests may be one condition necessary for a forest to have healthy trees, and thediversity and abundance of natural enemy species in the forest could be an appropriateassessment endpoint for measuring this condition. A conceptual model describing therelationships between an altered-lignin GM pine tree and potentially affectedinvertebrates and micro-organisms in a plantation forest was used to develop a set of riskhypotheses describing how the GM trees might affect each assessment endpoint. Becausepurified lignin does not represent the properties it imparts to wood, maximum hazard dosetests with non-target organisms, as are used to inform toxin risk assessment, cannot beconducted. Alternative experiments, based on current knowledge of the responses oforganisms to lignin, must be designed. A screening method was adapted and applied to adatabase of invertebrate species known to inhabit New Zealand pine forests to identify andprioritize non-target invertebrate species that could be used as experimental subjects forexamining these hypotheses. The screening model and its application are presented, alongwith a set of recommendations for pre-release tests with GM pines and potentially affectedinvertebrates and micro-organisms.