Uncontrolled clinical trials have shown that parenteral administration of GM-CSF reduces the frequency of chemotherapy-induced mucositis. The mechanism of this effect could be related to acceleration of haematopoiesis and/or increase in functional activation of WBC. We conducted a double-blind, placebo-controlled, dose ranging study of GM-CSF (mol-gramostim) mouthwash in patients with breast cancer during the first treatment cycle of a combination chemotherapy regimen which has historically produced dose-limiting (grade > or = 3) mucositis in approximately 39% of patients. Subjects were randomized to receive either placebo mouthwash (0.1 percent albumin) or one of four concentrations of GM-CSF mouthwash (0.01, 0.1, 1.0 or 10 mcg/ml). The primary endpoint was to evaluate the relationship between dose of GM-CSF mouthwash received and probability of grade > or = 3 mucositis using a logistic model. Solutions were administered four times daily starting within 24 hours of chemotherapy initiation and continuing until the end of the cycle (day 21). Mucositis was assessed on days 1-6, 10, 15 and 21. Day 6 plasma samples were assayed for GM-CSF. Forty-five patients were evaluable for response (nine per dosing group). A 42% risk (15/36) of mucositis grade > or = 3 was evident on day 15 in patients receiving GM-CSF compared to 2 of 9 patients on the placebo arm. No evidence of dose response was found by logistic regression. Five patients had a detectable plasma concentration of GM-CSF (56-209 pg/ml). A positive correlation between GM-CSF dose and leukocyte recovery was noted (P = 0.04).