We draw attention to an effect that has the potential to compromise measurements of atmospheric 14CO by the in situ production of 14C in air sample cylinders. We present first experimental determinations of the size of this effect using measurements made on cylinders containing an air sample and a CO/nitrogen calibration gas stored in laboratories in New Zealand from a few months to 11 years. These data are collated with measurements made on air samples stored at Scott Base, Antarctica to determine a mean in situ production rate for 14CO of 4.7 ¿ 0.5 molecules cm-3 yr-1 (STP) at sea level and in the latitude range 40--90¿ (geomagnetic) in a single floor laboratory. This rate is significant compared to observed annual cycles in 14CO (typically 5--10 14CO molecules cm-3) caused by OH variations and clearly has to be corrected for when 14CO data are used to determine atmospheric OH distributions.