In a series of studies, we evaluated the susceptibility of radioimmunoassays (RIA) for saliva cortisol to interference effects caused by oral stimulants used to facilitate saliva collection in studies with children. When added directly to saliva samples, oral stimulants (drink mix crystals) artificially inflated estimated cortisol concentrations. The magnitude of the interference effect was concentration-dependent and more pronounced for some stimulants and RIA procedures than for others. Analysis of samples collected using oral stimulants from child and adult participants confirmed stimulant interference as an extraneous source of variability in measured saliva cortisol. Associations between serum and saliva cortisol and between saliva cortisol and "behavioral" variables were attenuated by stimulant interference. A survey of six large child studies estimated interference effects, indexed by low sample pH, to be present in 14.7% of the 1,148 total saliva samples, or 2%-54% (M = 22%) of samples within each study. Recommendations to minimize the impact of stimulant interference in studies involving salivary cortisol in the context of child health and development are outlined.