The map of weight percent TOC of North Sea surface sediments was derived from geostatistical interpolation of TOC measurements gained from more than 3000 samples collected between 1980 and 2014. The data was compiled from various national and European data centers, project databases and the literature. The compiled dataset was statistically analysed to identify local extreme values and possible methodological artifacts. Kriging with external drift (KED) was then applied to predict the weight percent TOC at unsampled locations using the percentage sediment mud content as trend variable. 1. Datasets on weight percent TOC of bulk surface sediments (< 2 cm depth) were collected from national and European data centers and the literature. 2. Data were mapped according to the specified coordinates using WGS84 geodectic datum. 3. Samples located in the Kattegat area, the Orkney and Shetland archipelagos, the Wash, the back barrier tidal flats of the Wadden Sea Islands as well as to firths, fjords and estuaries were excluded from the dataset. 4. Zero measurements were excluded from the dataset 5. Local outliers were identified based on Cook's distance from a linear regression of spatially lagged (20 km) residuals and removed from the dataset if the cumulative probability of its Cook's D value exceeded 0.50 of the F distribution. 6. Spatial replicates (samples with identical geographic coordinates) were averaged. 7. At each location,the percentage mud content was sampled using information from https://geoportal.hzg.de/geoportal/catalog/search/resource/details.page?uuid={864FC61F-5C24-4433-9651-EC078E504E6A} 8. An experimental omnidirectional variogram for log10(TOC) with log10(mud) as a trend variable was calculated using Cressie's robust variogram estimator with a cut-off distance of 400 km ) and an increasing lag width of 5 to 30 km (a total of 22 bins). 9. The semi-variogram was approximated by an exponential variogram model to define the spatial autocorrelation structure required by kriging. 10. Using log10(mud) as trend variable, kriging with external drift (KED) was applied to locally predicted the weight percent TOC on a grid with a resolution of 1 x 1 nm. Unlike with variogram calculation and model fitting, outliers were included to account for the extreme values. 11. KED performance was validated by the hold out method using 70% of the data for training and the remaining 30% for testing. 12. Desired values of weight percent TOC were calculated by taking the anti-log of the kriging estimates. Note: All geostatistical computations were done with the R software environment (Version 3.3.2) using the “gstat” package (Pebesma, Computers & Geosciences, 30: 683-691)