The efficacy of two dentin-bonding agents (Gluma and Scotchbond Dual Cure) was investigated in vitro with or without saliva contamination of the dentin before or after application of the adhesive. The efficacy of the bonding agents was evaluated by a shear-bond strength test and a cavity test. When the dentin surface was contaminated, either before or after application of one of the bonding agents, the shear bond strength was reduced, but the difference between contaminated and non-contaminated specimens was not statistically significant. In the cavity test, the width and the extent of the marginal contraction gap were markedly increased when the dentin was contaminated with saliva before the bonding agent was applied. If the contamination happened after application of the adhesive, the efficacy of Gluma was further reduced, while that of Scotchbond was significantly improved. (This phenomenon has been investigated in detail in an accompanying paper.) The clinical consequence of saliva contamination seems to be that when Gluma is used as dentin-bonding agent, the contaminated area must be removed operatively, and then each of the various bonding procedures must be repeated.