When seven immunocompromised pateints developed invasive aspergillosis during construction at a hospital, new methods were performed to compare fungal isolates and a ease-control study was conducted to determine risks for infection. Typing of Aspergillus flavus with the use of restriction endonuclease analysis and restriction fragment length polymorphism using random amplified polymorphic DNA reactions to generate DNA probes revealed different patterns between isolates from two patients and a similar pattern among those from one patient, a health care worker, and an environmental source. Case patients were more likely than controls to have longer periods of hospitalization (median, 83vs.24 days;P< 0.01), neutropenia (median, 33vs.6 days;P< 0.05), and exposure to broad spectrum antimicrobials (median, 56vs.. 15 days;P= 0.08). No patients restricted to protected areas developed aspergillosis. Risk of exposure of immunocompromised patients to opportunistic organisms stirred up by construction activity may be decreased by admitting these patients to protected areas away from construction activity and by restricting traffic from construction sites to these areas. Although typing ofA. flavusisolates did not reveal a single type or source of organism responsible for infection, this method may facilitate epidemiologic inves