Current methods for vector-borne disease surveillance are limited by time and cost. To avoid human infections from emerging zoonotic diseases, it is important that the United States develop cost-effective surveillance systems for these diseases. This study examines the methodology used in the surveillance of a plague epizootic involving tree squirrels (Sciurus niger) in Denver Colorado, during the summer of 2007. A call-in centre for the public to report dead squirrels was used to direct animal carcass sampling. Staff used these reports to collect squirrel carcasses for the analysis of Yersinia pestis infection. This sampling protocol was analysed at the census tract level using Poisson regression to determine the relationship between higher call volumes in a census tract and the risk of a carcass in that tract testing positive for plague. Over-sampling owing to call volume-directed collection was accounted for by including the number of animals collected as the denominator in the model. The risk of finding an additional plague-positive animal increased as the call volume per census tract increased. The risk in the census tracts with >3 calls a month was significantly higher than that with three or less calls in a month. For tracts with 4–5 calls, the relative risk (RR) of an additional plague-positive carcass was 10.08 (95% CI 5.46–18.61); for tracts with 6–8 calls, the RR = 5.20 (2.93–9.20); for tracts with 9–11 calls, the RR = 12.80 (5.85–28.03) and tracts with >11 calls had RR = 35.41 (18.60–67.40). Overall, the call-in centre directed sampling increased the probability of locating plague-infected carcasses in the known Denver epizootic. Further studies are needed to determine the effectiveness of this methodology at monitoring large-scale zoonotic disease occurrence in the absence of a recognized epizootic.