A procedure has been developed for preparing 3- to 9-ℓ volumes of air under natural conditions with a known concentration of 222Rn to be used for calibrating radon systems. Air is passed into a plastic bag through a standard 226Ra solution (prepared by the U.S. National Bureau of Standards) contained in an emanation flask. This plastic bag retains 222Rn with little loss into or through the bag walls. The mean ratios of the 222Rn in the air at 2 and 7 days after filling to that immediately after filling were 0.992 ± 0.006 and 0.969 ± 0.008, which suggests a rate of radon loss into the bag of 0.4 ± 0.1%/day. The air from the bag was used to calibrate six Lucas chambers. Each chamber was calibrated 11 times with an average fractional standard error of the mean of 0.5%. This value is greater than the 0.2% expected from counting errors alone and suggests that the entire calibration procedure plus the counting system introduces a systematic standard deviation of 1.4% for each individual calibration and counting procedure. The bag and calibrated counters can also be used to determine the 226Ra and the 222Rn content of water. In addition, by replacing the air with other flush gases, calibration factors for gas mixtures other than air can be determined. The accuracy of these calibrations was verified by comparison with four methods, three of which are completely independent. The results by all four methods agree within ±1%.