Few studies have explicitly linked nutrient stoichiometry of catchments and lakes. We related N:P stoichiometry of stream exports from three catchments to that in downstream reservoirs varying in catchment land use. We also quantified heterocystous cyanobacteria and phytoplankton nutrient limitation. The N:P of stream exports varied greatly among catchments, being 18, 54 and 153 (molar) in the forested, mixed-use, and agricultural catchment, respectively. Total N:P in the mixed layers of reservoirs was less variable but showed a similar trend: 35, 52, 132. Phytoplankton in the forested lake were consistently N-limited, those in the agricultural lake were consistently P-limited, and those in the mixed-use lake showed a seasonal transition from P- to N-limitation. Heterocystous cyanobacteria were relatively more abundant in the forested lake than the agricultural lake. The N:P ratio of sediments buried in reservoirs varied with catchment input N:P, but differences were slight: sediment N:P was 4.3, 5.6, and 7.3 in the forested, mixed and agricultural lakes. Our results show congruence between the catchment export N:P, reservoir N:P, and phytoplankton N/P limitation. However, N:P stoichiometry of sediment retention was relatively insensitive to catchment stoichiometry, and suggests more efficient retention of P and/or greater loss of N in these systems.