"Kala-azar" (or Indian Visceral Leishmaniasis) is a vector-borne infectious disease affecting communities in tropical and subtropical areas of the world. Bihar, a state in India, has one of the highest prevalence and mortality reported levels of Kala-azar. Yet, the magnitude of the problem is difficult to assess because most cases are handled by private health providers who are not required to and do not report them to the Ministry of Health. The impact of underreporting using district-level reported incidence data from the state of Bihar is the main goal of this manuscript. We derive expressions for, and compute estimates of Kala-azar's reproduction numbers, an indirect measure of disease prevalence, and levels of underreporting for the 21 most affected districts of Bihar. The average reproduction number (number of secondary cases generated per infective) estimates for Bihar range from 1.3 (2003) to 2.1 (2005) with some districts' estimates with mean values lower than one. Model estimates (using available data and a model-derived expression) show that the proportion of underreported cases declined from an average of 88% in 2003 to 73% in 2005. However, eight districts in 2003 and five districts in 2005 had more than 90% levels of underreporting. Model estimates are used to generate underreporting adjusted incidence rates. The analysis finds that reported data misidentify four of the eight (2003) and three of the nine (2005) districts classified as high-risk. In fact, seven (2003) and five (2005) of the most affected Kala-azar districts had been classified as low-risk when only reported incidence data were used. Â\copyright 2009 Elsevier Ltd.