A network of in situ sensors and nutrient analyzers was deployed to measure nitrate, specific conductance (surrogate for chloride), and turbidity (surrogate for total suspended solids (TSS)) for 28 days in two urban streams near Minneapolis, MN. The primary objectives of the study were: (1) to determine the accuracy associated with quantifying pollutant loading using periodic discrete (i.e., grab) samples in comparison to in situ near real-time monitoring and (2) to identify pollutant sources. Within a highly impervious drainage area (>35%) the majority of pollutant load (>90% for nitrate, chloride, and TSS) was observed to be discharged in a small percentage of time (<20%). Consequently, periodic sampling is prone to underestimate pollutant loads. Additionally, when compared to loads based on near real-time sampling, average errors of 19-200% were associated with sampling 1-2 times a month. There are also limitations of periodic sampling with respect to pollutant source determination. Resulting implications with regard to total maximum daily load (TMDL) assessments are discussed.