The expense and ineffectiveness of drift-based insecticide aerosols to control dengue epidemics has led to suppression strategies based on eliminating larval breeding sites. With the notable but short-lived exceptions of Cuba and Singapore, these source reduction efforts have met with little documented success; failure has chiefly been attributed to inadequate participation of the communities involved. The present work attempts to estimate transmission thresholds for dengue based on an easily-derived statistic, the standing crop of Aedes aegypti pupae per person in the environment. We have developed these thresholds for use in the assessment of risk of transmission and to provide targets for the actual degree of suppression required to prevent or eliminate transmission in source reduction programs. The notion of thresholds is based on 2 concepts: the mass action principal-the course of an epidemic is dependent on the rate of contact between susceptible hosts and infectious vectors, and threshold theory-the introduction of a few infectious individuals into a community of susceptible individuals will not give rise to an outbreak unless the density of vectors exceeds a certain critical level. We use validated transmission models to estimate thresholds as a function of levels of pre-existing antibody levels in human populations, ambient air temperatures, and size and frequency of viral introduction. Threshold levels were estimated to range between about 0.5 and 1.5 Ae. aegypti pupae per person for ambient air temperatures of 28 degrees C and initial seroprevalences ranging between 0% to 67%. Surprisingly, the size of the viral introduction used in these studies, ranging between 1 and 12 infectious individuals per year, was not seen to significantly influence the magnitude of the threshold. From a control perspective, these results are not particularly encouraging. The ratio of Ae. aegypti pupae to human density has been observed in limited field studies to range between 0.3 and >60 in 25 sites in dengue-endemic or dengue-susceptible areas in the Caribbean, Central America, and Southeast Asia. If, for purposes of illustration, we assume an initial seroprevalence of 33%, the degree of suppression required to essentially eliminate the possibility of summertime transmission in Puerto Rico, Honduras, and Bangkok, Thailand was estimated to range between 10% and 83%; however in Mexico and Trinidad, reductions of >90% would be required. A clearer picture of the actual magnitude of the reductions required to eliminate the threat of transmission is provided by the ratio of the observed standing crop of Ae. aegypti pupae per person and the threshold. For example, in a site in Mayaguez, Puerto Rico, the ratio of observed and threshold was 1.7, meaning roughly that about 7 of every 17 breeding containers would have to be eliminated. For Reynosa, Mexico, with a ratio of approximately 10, 9 of every 10 containers would have to be eliminated. For sites in Trinidad with ratios averaging approximately 25, the elimination of 24 of every 25 would be required. With the exceptions of Cuba and Singapore, no published reports of sustained source reduction efforts have achieved anything near these levels of reductions in breeding containers. Practical advice on the use of thresholds is provided for operational control projects.