A proficiency test (PT) was organised for the detection of packaging materials in bakery by-products intended to be used as feed ingredients. Two blind samples of a size of 250 grams at spike levels of 50 mg and 250 mg per sample were distributed to 29 participants. Four participants analysed only a part of the sample material, leaving 25 data points eligible for further evaluation. The evaluation of the results was carried out according to the principles of the Standard ISO 17043:2010 and the Quality Guidelines for visual research, in the framework of individual spiking of the samples. This specific procedure was chosen to avoid large inhomogeneity of a general batch as basis for the production of the PT samples, with the consequence that usual statistics such as Z-scores could not be applied or needed modified interpretations. Results of this PT have been compared with the first PT in 2016 (n=22 participants, same spike levels). Altogether this provided the presence of four datasets. Blanks were not included in both PTs. The average recovery for the 50 mg level was 179%, and 105% for the 250 mg level in the 2019 PT. The maximum overestimations were 441% (50 mg) and 170% (250 mg), respectively. Symmetrical recovery intervals were chosen of 34%-166% for the 50 mg level and 66%-134% for the 250 mg level. A share of 80% of the results for the 250 mg sample was achieved within the limits of the recovery interval. A considerable number of overestimations was reported for the 50 mg level (12 participants, 48%) in the 2019 PT. Notwithstanding the limitations of statistical parameters, a higher precision in 2019 compared to the 2016 datasets can be concluded based on the smaller interval of the results, visible in the minimum-maximum ranges, in the 95% confidence intervals and in the lower standard deviations. In all cases the distributions are skewed to the right, shown by the underestimations in only one dataset, lack of outliers below the Lower Confidence Level and skewness values (much) higher than zero. The number of overestimations was the major cause of the lack of compliance with the uncertainty intervals. This might be due to either insufficient removal of water and fat from the selected particles of packaging material, and/or the selection of other particles mimicking the packaging material (specificity issue). The precise background of the overestimations needs further evaluation. False negatives were not reported in both the 2016 nor the 2019 version of the PT. The methods are applicable in the framework of enforcing a zero tolerance prohibition when transferring the observations to a compliant or non-compliant decision in relation to a threshold. Enforcement of an action limit could be possible from a level of 0.1% or higher in the view that the number of underestimations is very limited.Specificity as issue in the methods needs further investigation in order to document the precise cause of the deviations. Documentation for precise identification of packaging material needs to be developed.