The assessment of historical data is important to understand long-term changes in the marine environment. Whereas time series analyses based on monitoring data typically span one or two decades, this work aimed to integrate 40 years of monitoring and research data on polychlorinated biphenyls (PCBs) and metals in the Belgian Part of the North Sea (BPNS). Multiple challenges were encountered: sampling locations changed over time, different analytical methods were applied, different grain size fractions were analyzed, appropriate co-factors were not always analyzed, and measurement uncertainties were not always indicated. These issues hampered the use of readily available, highly standardized trend modeling approaches like those proposed by regional sea conventions such as OSPAR, named after the Oslo and Paris conventions.Therefore, we applied alternative approaches, allowing us to include most older historical data that have been obtained during the nineteen seventies and eighties. Our approach included reproducible and quality controlled procedures from data collection up to data assessment. It included spatial clustering, data normalization and parametric linear mixed effect modeling. A Ward hierarchical clustering was applied on recently obtained contaminant data, as the basis for a spatial division of the BPNS into five distinct areas with different contamination profiles. To minimize the risk of normalization errors for the metal data analyses, four normalization approaches were applied and mutually compared: granulometric and nickel (Ni) normalization, next to two hybrid normalization methods combining aluminum (Al) and iron (Fe) normalization. The long-term models revealed decreasing trends for most metals, except zinc (Zn) for which three out of four models showed increasing concentrations in all five zones of the BPNS. Offshore sediments contained the lowest normalized mercury (Hg) and cadmium (Cd) concentrations but high arsenic (As) concentrations. Trend analysis revealed a strong decrease in PCB ...