Bibliography: leaves 124-136. This thesis gives an overview of factors used in weed risk assessments and explores the disparity between the measured high accuracy rate of the weed risk assessment system (WRA) as implemented in Australia and the pessimistic assessments of some workers about the possibility of predicting the weed potential of plant species imported in the future. The accuracy of the WRA may not be as high as previously thought, and it varies with weed definition and taxonomic groups. Cluster analysis and comparative analysis by independent contrasts were employed to determine the value of individual biological and ecological questions on the WRA questionnaire. Results showed that some WRA questions could be deleted from the questionnaire and the scores for others weighted differently. The WRA is not a reliable predictor of weeds when it is considered in the context of the base-rate probability of an introduced plant becoming weedy in Australia. As a result a far greater number on non-weeds will be placed on the prohibited imported list than was initially expected.