Plasma fibronectin was measured in patients with breast cancer, colon cancer, and acute leukemia. In the patients with solid tumors, mean levels were significantly elevated above the mean level of age- and sex-matched normals whether the disease was thought to be metastatic or not (P less than 0.001). It did not make a difference whether the determinations were done prior to or during chemotherapy. Fibronectin was measured serially in eight hospitalized patients with leukemia during intensive induction chemotherapy. Normal concentrations were found prior to therapy. However, fibronectin concentration fell on the day following chemotherapy in nine of 12 episodes (P less than 0.05), and during sepsis in 13 of 13 episodes (P less than 0.001). Thus, the concentration was influenced by at least two factors: recent chemotherapy and sepsis. Because fibronectin concentration is sensitive to clinical events other than the status of the malignancy, it seems unsuitable as a tumor marker, at least as a single isolated measurement.