There are several methodological problems related to evaluating the impact that improved water supply, sanitation and health education projects have on public health. Fifteen studies of intervention programmes, one by the present authors, are discussed. Since non-intervention studies of water supply are difficult to assess, a valid study design must include an intervention programme, even though such studies are more costly and time-consuming. Detailed descriptions of programmes, study plans, and analytical methods are often lacking, and different studies reach greatly diverging conclusions. It may be proposed that studies based on "weak" methodology give a more positive impression of improved public health after improvement of water supply than do before-after studies using comparison groups.