Urban watersheds are significantly anthropogenically-altered landscapes. Most previous studies cover relatively short periods, without addressing concentrations, loads, and yields in relation to annual climate fluctuations, and datasets on Ag, Se, PBDEs, and PCDD/Fs are rare. Intensive storm-focused sampling and continuous turbidity monitoring were employed to quantify pollution at two locations in the Guadalupe River (California, USA). At a downstream location, we determined loads of suspended sediment (SS) for 14 yrs., mercury (HgT), PCBs, and total organic carbon (TOC) (8 yrs), total methylmercury (MeHgT) (6 yrs), nutrients, and trace elements including Ag and Se (3 yrs), DDTs, chlordanes, dieldrin, and PBDEs (2 yrs), and PCDD/Fs (1 yr). At an upstream location, we determined loads of SS for 4 yrs. and HgT, MeHgT, PCBs and PCDD/Fs for 1 yr. These data were compared to previous studies, climatically adjusted, and used to critically assess the use of small datasets for estimating annual average conditions. Concentrations and yields in the Guadalupe River appear to be atypical for total phosphorus, DDTs, dieldrin, HgT, MeHgT, Cr, Ni, and possibly Se due to local conditions. Other pollutants appear to be similar to other urban systems. On average, wet season flow varied by 6.5-fold and flow-weighted mean (FWM) concentrations varied 4.4-fold, with an average 7.1-fold difference between minimum and maximum annual loads. Loads for an average runoff year for each pollutant were usually less than the best estimate of long-term average. The arithmetic average of multiple years of load data or a FWM concentration combined with mean annual flow was also usually below the best estimate of long-term average load. Mean annual loads using sampled years were also less than the best estimate of long-term average by a mean of 2.2-fold. Climatic adjustment techniques are needed for computing estimates of long-term average annual loads.