High-frequency sampling of two major stream inflows to a large eutrophic lake (Lake Rotorua, New Zealand) was conducted to measure inputs of total suspended sediment (TSS), and fractions of nitrogen and phosphorus (P). A total of 17 rain events were sampled, including three during which both streams were simultaneously monitored to quantify how concentration–discharge (Q) relationships varied between catchments during similar hydrological conditions. Dissolved inorganic nitrogen (DIN) concentrations declined slightly during events, reflecting dilution of groundwater inputs by rainfall, whereas dissolved inorganic P (PO4–P) concentrations were variable and unrelated to Q, suggesting dynamic sorptive behaviour. Event loads of total nitrogen (TN) were predominantly DIN, which is available for immediate uptake by primary producers, whereas total phosphorus (TP) loads predominantly comprised particulate P (less labile). Positive correlations between Q and concentrations of TP (and to a lesser extent TN) reflected increased particulate nutrient concentrations at high flows. Consequently, load estimates based on hourly Q during storm events and concentrations of routine monthly samples (mostly base flow) under-estimated TN and TP loads by an average of 19% and 40% respectively. Hysteresis with Q was commonly observed and inclusion of hydrological variables that reflect Q history in regression models improved predictions of TN and TP concentrations. Lorenz curves describing the proportions of cumulative load versus cumulative time quantified temporal inequality in loading. In the two study streams, 50% of estimated two-year loads of TN, TP and TSS were transported in 202–207, 76–126 and 1–8 days respectively. This study quantifies how hydrological and landscape factors can interact to influence pollutant flux at the catchment scale and highlights the importance of including storm transfers in lake loading estimates.