Surveys in streams draining intensively farmed catchments can, during flood events, indicate differential time-concentration patterns between a bacterial health-risk indicator (E. coli) and a major zoonotic pathogen that it seeks to indicate-Campylobacter. The indicator's peak concentration at a monitoring station can arrive ahead of the flood peak (the pollutograph leads the hydrograph), whereas the peak pathogen concentration arrives with the flood peak (the pollutograph and hydrograph peaks coincide). In other cases the E. coli pollutograph can lag the hydrograph. These observations have generated the hypothesis that such behaviour reflects three different possible (predominant) sources of pathogens in the floodwater: (i) by sediment entrainment, (ii) via local land runoff, or (iii) from upstream releases (e.g., from dams, inflows, or upstream floods). A general theory for contaminants in idealized stream floods has been developed, considering all three sources, based on kinematic wave theory. It can explain the observed differential time-concentration patterns. The calculation procedures and associated results are intended to inform public policy, by identifying predominant pathogen sources and therefore helping to focus attention on the important delivery mechanisms. This will better inform quantitative health risk assessments for downstream water users (recreational uses, water supplies, food production and processing industries).