Conventional disease theory suggests that pathogens are unable to drive a host population to extinction if transmission of the pathogen is density-dependent. There are few empirical tests of this theory using manipulative experiments. We use an Ambystoma tigrinum – Ambystoma tigrinum virus (ATV) model to test the assumption that viral transmission is density-dependent.  We exposed susceptible A. tigrinum larvae to three different densities of infected larvae and three different frequencies of infected larvae for 24 hours. We then housed susceptible hosts individually for 28 days and monitored them for infection. The best predictor of infection was the density of infected hosts to which susceptible hosts were exposed (p=0.037); the density model had moderate predictive abilities (Area under the ROC curve = 0.64). Larvae in the highest density treatments also died sooner than larvae in lower density treatments (p < 0.001). ATV infection of susceptible larvae in populations that decline, recover, and do not go extinct is best modeled by the density of infected hosts. In contrast, density-independent transmission may be the most likely explanation for the enigmatic decline and even extinction of some amphibian populations resulting from emerging infectious disease.