We investigated the excitation temperature of atmospheric-pressure non-equilibrium (cold) plasma using a line-pair method. An atmospheric cold plasma was intermittently generated using a quartz tube, a rare gas, and a foil electrode by applying high-voltage alternating current. Because the plasma occurred intermittently, an interval appeared between each generated plasma bullet. We assessed the time-averaged effective intensity from the observed intensity at each wavelength measured using a spectrometer. When the exposure time of the spectrometer is longer than the observed period of the plasma emission light, the time-averaged effective intensity at each wavelength decreases because it includes the interval with no plasma emission light. The difference in intensity between wavelengths changes with frequency, because changing the frequency changes the interval between plasma bullets. Consequently, even if the plasma electron temperature does not depend on the frequency of the applied voltage, the excitation temperature estimated from the difference in intensity changes with the frequency. The plasma electron temperature can be estimated from the duty ratio of the observed period of plasma emission light, and we estimated the electron temperature in the helium and argon cold plasmas to be 1.0 eV.