Householders in a community-based programme to control dengue in El Progreso, Honduras, are being encouraged to improve the cleaning of the water-storage containers in which many of the vectors thrive. The objective of the present study was to develop an indicator of the change occurring in human behaviour. Traditional Aedes aegypti larval indices do not differentiate between containers in which all the immature stages are present and those which hold only first- and second-instar larvae. However, it is not essential to prevent all larval development to limit transmission of pathogens by the adults; if the Ae. aegypti in the containers only manage to develop to young larvae before the containers are cleaned, then control of the vector in these containers will be effective. In field trials, sampling of third- and fourth-instar larvae in washbasins by taking five dips (quick immersions to a standard depth) with a hand-held net was found to be sufficient for estimating the true population size of that same cohort. This sampling method was then included in a large-scale survey of households, conducted for programme monitoring. An index was then developed as a summary measure of the degree of infestation of a washbasin by Ae. aegypti. This index was the sum of four variables assessed in the survey: presence of any immature stages (larvae and/or pupae); presence of pupae; detection of third-fourth-instar larvae in a five-dip sample; and a log-transformation of the number of larvae recovered. Based on this new index, the 884 washbasins encountered in the survey were classified as infestation-free (76.2%), or with low-(6.7%), medium-(14.9%) or high-level (2.2%) infestation. Application of the same procedure to 240 drums encountered in the survey showed that 66.3% were infestation-free and 9.2%, 17.1% and 7.4% had low-, medium- and high-level infestations, respectively. Compared with the traditional indices, this new index should be more sensitive to changes in human behaviour resulting from a control programme exposure than a simple, dichotomous variable (i.e. positive/negative for presence of immature stages). The use of such an index could make the control programme more efficient, allowing the greatest efforts to be targeted at households that have medium-high levels of infestation.