Between November 1982 and July 1984, five patients at a 110-bed pediatric hospital were diagnosed with invasive filamentous fungal infection; three had invasive aspergillosis (IA) and two had invasive zygomycosis (IZ). All five had underlying hematologic malignancy (HM). In a case-control study, these five HM patients (cases) were compared to 10 autopsied HM patients without evidence of aspergillosis or zygomycosis (controls). Cases and controls did not differ in underlying disease or in the degree of immunosuppression, as measured by duration of granulocytopenia and number of platelet transfusions. However, case-patients were more likely than controls to have been hospitalized during the construction of a hospital addition (p less than 0.02, Fisher's exact test [FET]). Four (80%) of five HM patients autopsied during the period of construction had IA or IZ compared with one (5%) of 21 autopsied before construction began (p = 0.001, FET). These findings suggest that, in a population of comparably immunosuppressed patients, construction activity may represent an independent risk factor for IA or IZ. Hospitals caring for such patients should take precautions which minimize exposure of these patients to construction or renovation activity.