The aim is to introduce and characterize a new experimental demonstrative model contributing to the increase of measurement accuracy, in terms of minimum detectable activity (MDA) and background reduction, for the analysis of samples having low concentrations in tritium and radiocarbon on Quantulus 1220. The clue is related to the qualitative and quantitative differences between tritium and carbon‐14 inventories of the pulp used to manufacture the cups involved in noncatalytic combustion of samples by flame oxidation method. The quality of the experimental results depends on the temporal origin of the wood from which the pulp/cellulose was extracted/purified, the specific inventory contributing to the threshold level of the MDA for the beta‐emitting radionuclide to be investigated. Finally, the aim is to create and to use such a 100% old cellulose combustion cups for determining low concentrations of these radionuclides. It may be obtained by an adapted technology following the literature data and may be recommended for routine analyses of environmental samples coming from areas with no nuclear or minor nuclear influences, and also for low‐level biological samples. This first attempt resulted in improvement of measurement performances up to 400% for carbon‐14 and by approximately 50% for tritium.