Loss assessments are undertaken to support decisions about disaster mitigation. There is considerable pressure to use economic principles and to make such assessments a condition of funding for all mitigation. A fundamental underlying assumption is that loss assessments are accurate and comparable – and that this accuracy makes comparisons more valid. Unfortunately, it appears that this is not the case. A key question concerns whether loss assessments can be made accurate and comparable through improved knowledge and training – as implied by many critics of the approach – or whether the problems are inherent in the idea of loss assessment. Drawing primarily on Australian flood loss assessment work, these issues are examined. Results suggest that the uncertainties may be larger than generally acknowledged, that at least some are irreducible, and that comparisons may not be assisted by improved accuracy. The implication is that loss assessment methods should aim to make comparisons valid and reliable rather than chase unachievable precision.