A study of water quality variation in shallow protected springs in Kampala was undertaken over a 12-month period to assess the causes of microbiological contamination. The microbiological quality of water was assessed using thermotolerant coliforms and faecal streptococci. Sanitary inspections and hazard assessments were undertaken to identify faecal sources (hazards), contaminant pathways and contributory factors. Data were collected on rainfall and population as additional factors potentially exerting an influence on microbiological quality. Initial analysis of the data showed a significant relationship between median level of contamination and rainfall, in particular to short-term rainfall events. Total sanitary risk score showed a significant relationship with median level of contamination, but population density may be a confounding factor. The raw microbiological data were transformed into five water quality targets: <1 and < or =10 cfu 100ml(-1) for faecal streptococci; and <1, < or =10 and < or=50 cfu 100 ml(-1) for thermotolerant coliforms. The presence of individual risk factors as well as variables for rainfall and population density were analysed with respect to failure to meet these water quality targets using contingency tables. Logistic regression models were developed for each of the five water quality targets. The analysis strongly suggested that there is rapid recharge of the springs after rainfall and this leads to microbiological contamination. On-site sanitation was less important than other sources of faecal matter, which was consistent with a low sanitation coverage in the study area. The study suggested that improving sanitary completion and local environmental hygiene was more important than controlling on-site sanitation in improving the quality of these springs.