BACKGROUND It is unclear how much diphenhydramine (DPH) is toxic in humans. Previous dose-response studies have had conflicting results. Objective. We sought to evaluate DPH dose-response using a unique method that utilizes acetaminophen (APAP) serum concentrations to estimate DPH doses in patients ingesting APAP/DPH in a fixed-combination product. METHODS We retrospectively analyzed APAP/DPH-only exposures in patients 2-80 years of age using case data from 15 U.S. poison centers. DPH dose was extrapolated from measured serum APAP concentrations. A clinically significant response (CSR) was predefined in terms of eight specific manifestations (e.g., coma) that would warrant emergency department intervention. Nominal logistic regression was used to model the probability of each recorded manifestation across DPH dose ranges examining fits for mg, mg/kg, log10 mg, and log10 mg/kg DPH doses. The threshold value where patients reliably became symptomatic was determined by further examining receiver operating characteristic curves. RESULTS There were 509 cases that met inclusion criteria. Forty-five patients (9%) developed CSRs. A higher percentage of patients developed CSR at ≥ 7.5 mg/kg DPH and ≥1 g total DPH cutoff points (p < 0.05, Fisher's exact test). The best model for predicting the probability of CSR was a logistic fit of log(10) mg/kg dose (p < 0.05). By this model, for every 1 log(10) unit increase of mg/kg DPH dose, the odds of developing a CSR increased 47-fold (95% CI 17, 154). Receiver operating characteristic analyses showed a dose-related progression of symptoms. The cut-point with greatest sensitivity (98%) versus 1-specificity (57%) corresponded to an extrapolated mg/kg DPH dose of 8.2 mg/kg (95% CI 5.6, 10.5). CONCLUSIONS Our findings support the current American Association of Poison Control Centers' guideline recommendation to refer patients to the hospital for evaluation if they have ingested greater than or equal to 7.5 mg/kg of DPH.