For ion chambers with cavities open to the surrounding atmosphere, the response measured at a given temperature and pressure must be corrected using the standard temperature-pressure correction factor (P(TP)). A previous paper based solely on Monte Carlo simulations [D. J. La Russa and D. W. O. Rogers, Med. Phys. 33, 4590-4599 (2006)] pointed out the shortcomings of the P(TP) correction factor when used to correct the response of non-air-equivalent chambers for low-energy x-ray beams. This work presents the results of several experiments that corroborate these calculations for a number of ion chambers. Monte Carlo simulations of the experimental setup revealed additional insight into the various factors affecting the extent of the breakdown of P(TP), including the effect of impurities and the sensitivity to chamber dimensions. For an unfiltered 60 kV beam, the P(TP)-corrected response of an NE 2571 ion chamber measured at 0.7 atm was 2.5% below the response measured at reference conditions. In general, Monte Carlo simulations of the experimental setup using EGSnrc were within 0.5% of measured values. EGSnrc-calculated values of air kerma calibration coefficients (N(K)) at low x-ray energies are also provided as a means of estimating the level of impurities in the chambers investigated. Calculated values of N(K) normalized to the value measured for a 250 kV beam were obtained for three chambers and were within 1% of experiment with one exception, the Exradin A12 in a 50 kV beam.