The sampling performance of C-18 cartridges coated with DNPH has been studied for twenty four C-1-C-9 carbonyls in experiments involving sampling of parts per billion levels of carbonyls in urban air, indoor air and laboratory experiments. The cartridge background carbonyl content in thirty six batches of cartridges averaged 85, 137 and 155 nanogram/cartridge for formaldehyde, acetaldehyde and acetone, respectively, and was below analytical detection for all other carbonyls. Carbonyl-DNPH derivative recovery from the cartridge was complete in the first elution with 2 mt acetonitrile, and this for twenty four carbonyls at concentrations of 0.02-73 mu g carbonyl/cartridge. Studies carried out using two cartridges in series showed no breakthrough, for the sixteen carbonyls tested, at concentrations of 0.10-49 mu g carbonyl/cartridge and volumes of air sampled = 6-370 L. Average relative standard deviations (RSD) for replicate analyses were 0.20-13.2% for twenty one carbonyls. Average RSD for co-located samples were 0.9-16.2% for eighteen carbonyls. Comparison of RSD for replicates and RSD for co-located samples for thirteen carbonyls indicated that the overall method precision was limited by sampling precision rather than by analytical precision.