Interlaboratory comparsion exercises on the determination of polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDDs/PCDFs) and dioxin-like polychlorinated biphenyls (PCBs) in three natural foodstuffs have been performed annually since 2000 by the Norwegian Institute of Public Health. Up to 77 laboratories from 24 countries world-wide have reported results. The data have been statistically analysed in order to assess the comparability of both the reported individual congener concentrations and the calculated toxic equivalents (TEQ). Due to the low concentrations of PCDDs/PCDFs and dioxin-like PCBs in food, the treatment of congeners reported as "not detected" and the identification of outliers, influence the estimation of the consensus values considerably. Based on the experiences from the first two rounds, an approach for calculation of the consensus values was established, resulting in a non-skewed distribution of data. The relative standard deviation of consensus TEQs was 13-32% for PCDDs/PCDFs and 11-45% for PCBs. Z-scores were calculated for PCDD/PCDF-TEQs using a target value of +/-20% for the deviation. For food samples with a PCDD/PCDF contamination of more than half of the EU maximum level, 63-87% of the participants achieved Z-scores +/-1, while for low contaminated foodstuffs about 60% of the participants had Z-scores of +/-2. After four rounds of interlaboratory comparisons it can be concluded that there is a good comparability of the analytical data for dioxins and dioxin-like PCBs in food at higher contamination levels. However, the laboratories need to improve their performance when determining levels in food with low contamination of dioxin-like compounds, due to the increasing demand for analyses of such food items. The importance of determining dioxin-like PCBs is demonstrated by their large contribution to the total TEQ, especially in food from the marine environment.