Traditional methods of monitoring gray wolves (Canis lupus) are expensive and invasive and require extensive efforts to capture individual animals. Noninvasive genetic sampling (NGS) is an alternative method that can provide data to answer management questions and complement already-existing methods. In a 2-year study, we tested this approach for Idaho gray wolves in areas of known high and low wolf density. To focus sampling efforts across a large study area and increase our chances of detecting reproductive packs, we visited 964 areas with landscape characteristics similar to known wolf rendezvous sites. We collected scat or hair samples from 20% of sites and identified 122 wolves, using 8–9 microsatellite loci. We used the minimum count of wolves to accurately detect known differences in wolf density. Maximum likelihood and Bayesian single-session population estimators performed similarly and accurately estimated the population size, compared with a radiotelemetry population estimate, in both years, and an average of 1.7 captures per individual were necessary for achieving accurate population estimates. Subsampling scenarios revealed that both scat and hair samples were important for achieving accurate population estimates, but visiting 75% and 50% of the sites still gave reasonable estimates and reduced costs. Our research provides managers with an efficient and accurate method for monitoring high-density and low-density wolf populations in remote areas.