During the past two decades, health interview surveys have become an increasingly common source of information about current morbidity patterns and utilization of health services in developing countries. This study describes a recent effort to enhance the utility of these surveys by incorporating a calendar format. Methods A calendar of morbidity and treatment behaviour during the 2-week period prior to interview was implemented in the Guatemalan Survey of Family Health (EGSF), a large-scale sample survey that was fielded in 60 communities in rural Guatemala in 1995. A total of 2872 women aged 18-35 were interviewed and provided information on 3193 children born since 1990. Results The EGSF calendar data provide estimates of diarrhoeal illness that are consistent with those obtained from more conventional questionnaire designs. However, in contrast to conventional health survey questions, these calendar data: (1) permit a much more complete evaluation of the accuracy of reporting; and (2) offer a richer and more complex description of child illness and treatment behaviour. For example, the results demonstrate that even the preferred 2-week recall period suffers from underreporting of diarrhoeal illness, that the majority of children with diarrhoea experience at least one additional symptom, and that mothers assess severity of diarrhoea from the type and number of accompanying symptoms. Conclusions The findings indicate that additional implementation and evaluation of calendar formats is warranted in order to provide the most useful and accurate data possible at relatively low cost