To determine the significance of Aspergillus species isolated from sputum or other respiratory secretions with respect to the diagnosis of invasive pulmonary aspergillosis, the clinical records and radiographs of all patients whose respiratory secretion cultures yielded an Aspergillus species between 1972 and 1978 were reviewed. All known predispositions to invasive aspergillosis, e.g., presence of cancer or granulocytopenia, and therapy with corticosteroids, antibiotics, and cytotoxic drugs, were significantly associated with proven or probable invasive pulmonary aspergillosis. Most notable were patients with acute leukemia and granulocytopenia. Prolonged duration of hospitalization between initial isolation (greater than 2 weeks) and multiple isolates (greater than three isolates) were also significantly associated with a high frequency of proven or probable disease. Isolation of A. niger was only rarely associated with proven or probable disease (one of eight patients). The isolation of A. fumigatus and A. flavus from respiratory secretions does not usually represent laboratory contamination and must be interpreted in the light of known predispositions to aspergillosis. In some situations, e.g., granulocytopenic patients with acute leukemia, even a single isolation carries a high likelihood of invasive aspergillosis.