Background: Rates of communicable disease among young children are considerably higher in Israel than in other western countries. Strategies for reducing the disease rates are needed. Purpose: The goal of the Jerusalem Handwashing Study (JHS) was to evaluate a preschool-based hygiene programme aimed at reducing illness absenteeism. Methods: The trial employed cluster randomization of preschools to intervention or control. The intervention programme was multifaceted, including various educational activities and environmental changes. The control group received the programme at the end of the study. A supplementary home component was tested using the innovative design strategy of an embedded individually-randomized trial. All-cause absenteeism and illness-related absenteeism was assessed by daily phone calls to the teachers, supplemented with phone calls to the parents to identify the cause of the absence. The primary endpoint was illness-related absenteeism. In an attempt to avoid bias, survey staff were blinded to the study design and the main outcome as well as intervention status. Validity checks were incorporated to assess the accuracy of educator absenteeism reports and were analysed for differences between the study arms. Observation of handwashing behaviour allowed assessment of behavioural change in the intervention group and spontaneous handwashing changes in the control group. Results: Cluster randomization with delayed implementation in the control group was a feasible and efficient strategy. The individually-randomized embedded subtrial proved to be an efficient way to test a supplemental intervention component, and is particularly well-suited to programmes run in educational or other group settings, including workplaces, clinics, and community centres. Limitations: The trial design did not permit analysis of the intervention effect on illness. Conclusions: The techniques used in this trial made it possible for a rigorous study of a complex community-based health intervention to be carried out successfully. They should prove helpful to future researchers.