The concept of nutrient limitation, as developed in agriculture, applies well to wild plants grown under controlled conditions, although plants adapted to infertile soils are less responsive to nutrient addition than are most crop species. There are serious difficulties in transferring the concept of nutrient limitation directly to plant communities, however, because (1) in comparing communities with different dominant species, the species characteristic of nutrient-rich sites are inherently more responsive to nutrient supply and may be more strongly nutrient-limited than species in low-nutrient sites, and (2) ecosystem-level feedback complicates the analysis of experiments involving fertilization. We suggest that nutrient limitation in communities can be best measured by assessing the plants' response to large nutrient additions that are sufficient to saturate chemical and microbial immobilization processes and still meet plant nutrient requirements. The magnitude of community-level nutrient limitation is highly sensitive to the potential growth rate of component species; it may be greatest in sites of intermediate fertility.