PURPOSE OF REVIEW: There is recognition that respiratory sensitization is an occupational hazard of high concern. Despite international regulatory requirements there is no established protocol for the efficient prospective identification of chemical respiratory sensitizers. We review the predictive behaviour of available methods and suggest a possible high-throughput protocol. RECENT FINDINGS: Animal or in-vitro tests specific to respiratory exposure and resulting in direct asthma-related outcomes have not been developed, although the use of a local lymph node assay originally designed for skin sensitization has been advocated in a respiratory context. Various methods have been used to develop quantitative structure-activity relationship models for prediction of low-molecular-weight organic chemical respiratory sensitizers. The estimated negative predictive value for all of the published models is 1, but their differences in positive predictive value can be exploited. SUMMARY: The most pragmatic as well as valid approach for screening large numbers of industrial chemicals for respiratory sensitization hazard is likely to consist of an algorithm starting with quantitative structure-activity relationship models. Further corroboration from animal or human data, however, may be required for chemicals with a positive result by quantitative structure-activity relationship.