Rivers have been increasingly polluted by nitrogen (N) and phosphorus (P) over the past century, and this trend may continue due to human activities under socio-economic development and climate change. This study aims to identify dark and bright spots of N and P deliveries into rivers with two large-scale water quality models: MARINA-Multi and IMAGE-DGNM. We quantified sources of these nutrients across 1303 sub-basins worldwide, focusing on both a historical period (2010) and a future projection (2050) using SSP1 (sustainability-driven with low global warming) and SSP5 (economically-driven with high global warming). Dark or bright spots were defined as sub-basins where N or P deliveries into rivers exceed or non-exceed the critical boundaries, above or below which pollution issues such as eutrophication may occur or not occur. Results showed high agreements between the two models on the dark and bright spots for N (at least 81% of sub-basins) and P (at least 74% of sub-basins), in both periods. In general, more bright spots were identified for N than for P, implying that N loads in many sub-basins worldwide are expected to stay below the critical boundaries in 2050 scenarios. In contrast, more dark spots for P imply that P loads in rivers are expected to exceed the critical boundaries for more than 50% of the global land area in 2050. Globally, the two models agreed that agriculture is the dominant source of nutrient pollution in more than 77% of the sub-basins for N, and more than 60% for P. The largest disagreement between the two models is on the contribution from human waste, which is linked to variations in input data and modeling approaches. Our study provides insights that open opportunities for further model intercomparisons to support effective water pollution control measures.