Introduction: The research was performed in two rural sites in Thailand and Laos. The Diarrhea and Dengue (DIADEN) group from the Norwegian University of Life Sciences facilitated the fieldwork. In the Thai site, rainwater was the main drinking water source, whereas in Lao site, unprotected dug well was predominant in dry season and was replaced by rainwater in wet season. Drinking water was largely consumed untreated. Combined with varying sanitation service and hygiene behavior, it was assumed that the safety of drinking water might be compromised. Problem definition: The global monitoring of the MDG target to halve the population without safe drinking water access is performed by the Joint Monitoring Programme (JMP). The JMP uses technological classification (improved or unimproved sources) as monitoring indicator for water safety. This has been criticized for being inadequate in representing the actual conditions in water provision. The core argument is that drinking water quality at improved sources might not necessarily be free from pathogens. In addition, drinking water that is safe at the sources might as well be recontaminated through various water handling practices, particularly where manual collection, transport and storage are common. Thus, in this study it is proposed that the technological classification is refined through the use of a semi-quantitative hazard classification system. Research: In the first part, the research was focused on the identification and classification of the hazards found at the various water sources and the hazards related to sanitation facility and hygiene behavior. This was achieved through sanitary inspection, household questionnaire and spot observation carried out in about two weeks time. In the second part, water samples were collected twice at selected sources and household drinking containers. Samples were tested for E. coli as faecal indicator organisms, using the standard Colisure/Ouanti-tray 2000 method from IDEXX. The water quality data were then used to validate the hazard classification systems. Simultaneously, a new E. coli enumeration method, the Compartment Bag Test (CBT) by University of North Carolina, was used in parallel to investigate the comparability of the two tests. Results: 1. The first investigation found that the hazards in Laos were more severe than in Thailand. 2. Water quality of the rainwater source (improved) was variably contaminated. The unprotected dug well (unimproved) was grossly contaminated. 3. Water quality at the household containers was also variable. Overall, there was significant deterioration of water quality from sources to households that can be attributed to household water handling practices (e.g. extraction methods, cleanliness, treatment). 4. The hazard classification system for water sources in Laos was moderately correlated with water quality data and was not significantly correlated for Thailand. The sanitation and hygiene hazards were not significantly correlated in both sites. 5. The CBT data was significantly different from the Colisure/Quanti-tray 2000 data. Conclusions and recommendations: The JMP indicator overestimates water safety both at sources and at households. The hazard classification system, if refined further with weighted-scoring and long-term water quality data, might enable a more accurate representation of the actual conditions. Correlations with health impacts are also recommended. Furthermore, manual water handling practices compromises water quality and it is thus advised to eliminate hand-water contact as much as possible.