Eutrophication mitigation is an ongoing priority for aquatic ecosystems. However, the current eutrophication control strategies (phosphorus (P) and/or nitrogen (N)) are guided mainly by nutrient addition experiments in small waters without encompassing all in-lake biogeochemical processes that are associated largely with lake morphological characteristics. Here, we use a global lake data set (573 lakes) to show that the relative roles of N vs P in affecting eutrophication are underpinned by water depth. Mean depth and maximum depth relative to mixing depth were used to distinguish shallow (mixing depth > maximum depth), deep (mixing depth < mean depth), and transitional (mean depth ≤ mixing depth ≤ maximum depth) lakes in this study. TN/TP ratio (by mass) was used as an indicator of potential lake nutrient limitation, i.e., N only limitation if N/P < 9, N + P colimitation if 9 ≤ N/P < 22.6, and P only limitation if N/P ≥ 22.6. The results show that eutrophication is favored in shallow lakes, frequently (66.2%) with N limitation while P limitation predominated (94.4%) in most lakes but especially in deep ones. The importance of N limitation increases but P limitation decreases with lake trophic status while N and P colimitation occurs primarily (59.4%) in eutrophic lakes. These results demonstrate that phosphorus reduction can mitigate eutrophication in most large lakes but a dual N and P reduction may be needed in eutrophic lakes, especially in shallow ones (or bays). Our analysis helps clarify the long debate over whether N, P, or both control primary production. While these results imply that more resources be invested in nitrogen management, given the high costs of nitrogen pollution reduction, more comprehensive results from carefully designed experiments at different scales are needed to further verify this modification of the existing eutrophication mitigation paradigm.