Limits on the levels of selected mainstream smoke constituents per milligram of nicotine (nicotine ratio) have been proposed as an option for regulatory compliance of cigarettes. However, these limits must come with tolerances in order to avoid misleading conclusions on certain brands. To establish tolerances, it is essential to know the accuracy of the applied methods and for that, the widely used method is to perform a collaborative study in order to determinate the repeatability and reproducibility. This paper describes the different collaborative study processes used for accuracy measurement according to the dataset (raw data, ratios with or without paired measurement). Use of unpaired measurements for computing ratios highlights the difficulties of repeatability and reproducibility assessment. Indeed, due to analytical constraints the measurements of smoke components and smoke nicotine are not always carried out simultaneously (unpaired measurements), therefore the ISO 5725 statistical approach can be inappropriate. For example, accuracy of NNN yield per milligram of nicotine (from a recent CORESTA Special Analytes Sub-Group joint experiment) calculated using the ratios generated with permutation strategies, showed the wide range of estimated repeatability and reproducibility: 112% and 77%, respectively under ISO smoking regime for the 1R5F cigarette.

Therefore, as the accuracy of nicotine ratio expression, with unpaired measurement, cannot be clearly determined using the basic method for the determination of repeatability and reproducibility defined by ISO standard, some strategies will be suggested to approximate these uncertainties.