Effective monitoring of wildlife populations requires detailed ecological background and reliable data sources. Noninvasive DNA sampling can be a powerful tool for wildlife ecological studies as it can be deployed over large areas and reliably detect multiple species. Although the feasibility of employing noninvasive DNA for ecological studies has been well documented, few studies have used this technique in Iran. In this study, I evaluated the feasibility of performing species-level identification of sympatric carnivore species, using faecal samples collected in the wild. From June to October 2012, I collected 174 faecal samples from six protected areas in the Iranian Caucasus. First, I used morphological characteristics for field identification of each species. Then I used a short cytochrome b segment to identify the species of each scat. Using confusion matrices, I evaluated the accuracy in field identification for each species. A total of six carnivore species were successfully identified from 67. 2% of all samples obtained, including the brown bear (Ursus arctos), wildcat (Felis silvestris/lybica), grey wolf (Canis lupus), red fox (Vulpes vulpes), stone marten (Martes foina) and Eurasian otter (Lutra lutra). The lowest accurate identification of carnivore scats (true positive rates) were for the wildcat and Eurasian otter (0%). In contrast, accuracy for brown bear scat field identification was high (94%). I found a high false-positive rate for stone marten, indicating samplers often incorrectly assigned its scats to the red fox. This pilot study shows that DNA obtained from faecal samples works efficiently for surveying Iranian carnivores.