Separate models correlating natural gas (NG) consumption to climate have been developed for the residential and commercial sectors of the 50 U.S. states. The models relate a population-weighted average temperature to state per capita NG consumption on a monthly basis. The majority of the models have Pearson correlation coefficients greater than 0.90 supporting the use of temperature as the sole independent parameter. The sensitivities of the models to a 1 degrees C increase in temperature, are compared for each state and the monthly sensitivity to climate integrated over the entire U.S. is investigated for a range of temperature perturbations. The predicted impact of a 1 degrees C increase in mean monthly temperature on U.S. consumption is an 8.1% decrease in the residential sector and a 5.9% decrease in the commercial sector. In terms of the net consumption normalized over the study period (1984-1993) this corresponds to a 111.8 TWh decrease in the residential sector and a 47.0 TWh decrease in the commercial sector. The largest change for a single month occurs in January when consumption would decrease 19.7 TWh in the residential sector and 7.4 TWh in the commercial sector.