Climate-related extreme weather events can result in the loss of drinking water access. We assessed the relative vulnerability of 3143 United States (U.S.) counties to loss of drinking water access due to droughts, floods, and cyclones. Five vulnerability assessment models from the literature were compared, each differing in the aggregation method used to combine the three determinants of vulnerability (V) -exposure (E), sensitivity (S), and adaptive capacity (AC). Exposure scores were calculated using historical occurrence data, sensitivity scores were determined from the intrinsic resilience of the drinking water technologies, and adaptive capacity scores were calculated from nine socioeconomic indicators. Our results showed that models V=E+S+AC and V=E+S-AC were the same, as were models V= E×S×AC and V=E×S÷AC. Between these two model forms (form 1: V=E+S+AC and V= E+S-AC; form 2: V=E×S×AC and V=E×S÷AC), scores from one model form could be used to predict scores from the second model form, with R-squared values ranging from 0.61 to 0.82 depending on the extreme weather event type. A fifth model, V=(E-AC)×S was not found to correlate with any of the other four models. We used V=E+S+AC as our reference model as this resulted in a more uniform distribution of counties in each of the five intervals of vulnerability. Comparing the vulnerability scores identified the counties with greatest vulnerability to losing access to drinking water due to floods, droughts, and cyclones. Our results can Climatic Change (2015) 133:665-679