摘要: Abstract Reliable quantitative risk assessment requires the use of meaningful statistical data andwell validated models for their interpretation. In certain domains neither nor thevalidated are available. Nevertheless, there may still be a requirement to generatea characterisation risk, presented in form which reflects reliability andaccuracy One such domain is potential carcino-genic chemical compounds, and many others. Work described isunderway provide sound framework qualitative using acomputer model “ argumentation ”. This work being applied development ofcomputer-based support carcinogenic risk. 1. Introduction paper describes under way develop methods riskassessment. A specific focus this computer-based assistant theassessment compounds. However, approachshould have much wider application.For technologies, as electronic systems, civil engineering structures mechanicalsystems, established available making precise reliable estimates ofthe likelihood system failure. Significant quantities historical on failure rates ofstandard components available, example, or it possible generate reliablesimulations behaviour. other areas, however, position less satisfactory. Forexample, associated with compounds used foods agricul-ture, knowledge nature biological action complex chemicals so sparse thatthere considerable doubt about validity basis formulationof policy. The report Department Health’s Committee Carcinogenicity Chemicalsin Food (Carter, 1992) concludes:The committee does not routine chemi-cal carcinogens. because present validated, often based onincomplete inappropriate data, derived more from mathematical assumptions thanfrom mechanisms and, at least present, demonstrate dis-turbingly wide variation depending adopted.This conclusion echoes feelings professionals, managers policy makers otherfields. needed apply heldto simplistic, precision they offer illusory.