Underwater noise pollution from shipping is globally pervasive and has a range of adverse impacts on species which depend on sound, including marine mammals, sea turtles, fish, and many invertebrates. International bodies including United Nations agencies, the Arctic Council, and the European Union are beginning to address the issue at the policy level, but better evidence is needed to map levels of underwater noise pollution and the potential benefits of management measures such as ship-quieting regulations. Crucially, corroboration of noise maps with field measurements is presently lacking, which undermines confidence in their application to policymaking. We construct a computational model of underwater noise levels in the Northeast Atlantic using Automatic Identification System (AIS) ship-tracking data, wind speed data, and other environmental parameters, and validate this model against field measurements at 4 sites in the North Sea. Overall, model predictions of the median sound level were within ±3 dB for 93% of the field measurements for one-third octave frequency bands in the range 125 Hz-5 kHz. Areas with median noise levels exceeding 120 dB re 1 μPa and 20 dB above modelled natural background sound were predicted to occur in the Dover Strait, the Norwegian trench, near to several major ports, and around offshore infrastructure sites in the North Sea. To our knowledge, this is the first study to quantitatively validate large-scale modelled noise maps with field measurements at multiple sites. Further validation will increase confidence in deeper waters and during winter months. Our results highlight areas where anthropogenic pressure from shipping noise is greatest and will inform the management of shipping noise in the Northeast Atlantic. The good agreement between measurements and model gives confidence that models of shipping noise can be used to inform future policy and management decisions to address shipping noise pollution.