An ELISA for the detection and measurement of Aspergillus antigenaemia has been developed and evaluated by examining sera submitted over a 12-month period from immunocompromised patients with a likelihood of invasive aspergillosis. Results from proven cases of invasive aspergillosis confirmed at post-mortem and specimens from individuals with suspected disease showed that tests on single serum samples were often negative. Multiple specimens from the same patient greatly increased the frequency of detection. Repeated monitoring of sera from a single patient showed wide fluctuations in antigen level, which was considered to be due partly to the medical regimen to which the patient was subject. Control sera from healthy laboratory personnel were consistently negative, but a number of 'at-risk' patients without other evidence of invasive aspergillosis sometimes had low amounts of antigen. Concentrations of Aspergillus antigen of 100 ng ml-1 or higher were considered to be strongly suggestive of fungal invasion.