Accelerator mass spectrometry (AMS) measurements of the radiocarbon content in very old samples are often challenging and carry large relative uncertainties due to possible contaminations acquired during the preparation and storage steps. In case of such old samples, the natural surrounding levels of 14C from gases in the atmosphere, which may well be the source of contamination among others, are 2–3 orders of magnitude higher than the samples themselves. Hence, serious efforts are taken during the preparation steps to have the samples pristine until measurements are performed. As samples often have to be temporarily stored until AMS measurements can be performed, storage conditions also become extremely crucial. Here we describe an assessment of this process of contamination in background AMS samples. Samples, both as pressed graphite (on AMS targets) and graphite powder, were stored in various storage conditions (CO2-spiked air) to investigate the extent of contamination. The experiments clearly show that the pressed targets are more vulnerable to contamination than the unpressed graphite. Experiments conducted with enriched CO2-spiked laboratory air also reveal that the contaminating carbon is not only limited to the target surface but also penetrates into the matrix. A combination of measurements on understanding the chemical nature of the graphitization product, combined with long-available knowledge on “adventitious carbon” from the surface science community, brought us to the conclusion that contamination is to a certain extent inevitable. However, it can be minimized, and should be dealt with by sputter-cleaning the samples individually before the actual measurement.