Acute topical treatment of human skin with retinoic acid (RA) results in a pleiotropic response, some aspects of which are mimicked by non-specific irritants. To identify reliable cutaneous markers of retinoid action, it is important to determine which aspects of this response are specifically due to the presence of RA. We have previously demonstrated a rapid and pronounced increase in steady-state cellular RA-binding protein II (CRABP-II)mRNA levels after topical RA treatment. Here we characterize the dose dependence and kinetics of this response, and compare the effects of a well-known irritant, sodium dodecylsulfate (SDS), to those of RA and its vehicle. The induction of CRABP-II mRNA in response to 0.1% RA cream was maximal by 16 h (elevenfold relative to untreated skin), and persisted at near-maximal levels (eight-fold) for up to 4 d. RA was potent in eliciting this response, as approximately half-maximal stimulation was observed after 16 h of treatment with 0.001% RA. Treatment for 4 d with 0.1% RA cream versus 2% SDS in RA vehicle resulted in nearly identical levels of cutaneous erythema, spongiosis, and epidermal thickening. However, the CRABP-II mRNA response to 2% SDS was no greater than that observed in response to vehicle alone (2.9 times relative to occluded skin control at 4 d). SDS also had no effect upon either CRABP-II or RAR-beta mRNA levels in quiescent human dermal fibroblasts in vitro, whereas RA elicited both responses at 1000-times lower concentrations than SDS. Taken together, these data identify the CRABP-II mRNA response as a reliable, rapid, and selective marker for retinoid activity in human skin.