The theoretical underpinnings of the assessment of invasive alien species impacts need to be improved. At present most approaches are unreliable to quantify impact at regional scales and do not allow for comparison of different invasive species. There are four basic problems that need to be addressed: (1) Some impacted ecosystem traits are spatially not additive; (2) invader effects may increase non-linearly with abundance or there may be effect thresholds impairing estimates of linear impact models; (3) the abundance and impact of alien species will often co-vary with environmental variation; and (4) the total invaded range is an inappropriate measure for quantifying regional impact because the habitat area available for invasion can vary markedly among invasive species. Mathematical models and empirical data using an invasive alien plant species (Heracleum mantegazzianum) indicate that ignoring these issues leads to impact estimates almost an order of magnitude from the real values. Thus, we propose a habitatsensitive formula for regional impact assessment that is unaffected by non-linearity. Furthermore, we make some statistical suggestions on how to assess invader effects properly and we discuss the quantification of the invaded range. These improvements are crucial for impact assessment with the overall aim of prioritizing management of invasive species.