The effects of the adjuvant QS-21 in various formulations on immediate pain on injection after intramuscular injection were evaluated in three Phase I clinical trials in healthy adults. Each trial was designed as a double-blind, randomized, four-way or five-way cross-over study with each subject acting as his/her own control. In the first trial, four formulations designed to evaluate the effect of QS-21 or pH (over a range of 6--7.2) were evaluated: phosphate-buffered saline at pH 6.0 or 7.2, and 50 microg of QS-21 in phosphate-buffered saline at pH 6.0 or 7.2. Thirty-three volunteers received each of the four intramuscular injections in random order separated by approximately 1 week. The volunteers assessed the immediate injection pain from 0 to 10 (none to most pain). The data indicate that the presence of QS-21, but not pH, is associated with transient injection site pain. The second trial, which utilized the same design as the first trial, evaluated formulations of QS-21 in various excipients. Fifteen volunteers received phosphate-buffered saline, QS-21/PBS, QS-21/aluminum hydroxide, and QS-21/4 mg/ml of polysorbate 80. Polysorbate 80, but not aluminum hydroxide, reduced the mean pain score compared to QS-21/PBS. The third trial evaluated formulations of QS-21 in additional excipients. Fifteen volunteers received aluminum hydroxide (without QS-21), QS-21/PBS, QS-21/0.72% benzyl alcohol, QS-21/30 mg/ml of hydroxypropyl-beta-cyclodextrin, and QS-21/8-mg/ml of polysorbate 80. Benzyl alcohol, cyclodextrin, and the higher concentration of polysorbate 80 reduced the pain scores associated with QS-21. Hence, QS-21 is associated with injection pain in simple buffer formulations, but it is possible to improve the acceptability of QS-21-containing formulations through reformulation with certain excipients.