Large carnivores like lynx and wolves have recolonised parts of their historic range in central Europe. Lynx occurrences in Germany and Switzerland are recovering rather hesitantly, while wolf occurrences are steadily increasing. Due to legal norms and societal interest, the development of lynx and wolf occurrences is being observed in the framework of monitoring activities. Camera traps and genetic analyses of non-invasively collected material like scats are common monitoring methods. The latter is hard to implement for lynx, as sample material is difficult to find. The present work examines to what extend trained scat detection dogs are suitable for monitoring lynx and wolves. For this purpose, the use of scat detection dogs was compared with the application of camera traps as well as hair traps (lynx) and devices for recording howling events (wolf). For the lynx study two detection dog teams were simultaneously deployed in 44 grid cells of 4 km2 each in the German Bavarian Forest National Park for four weeks in 2017. Each grid cell additionally contained one camera and one hair trap. Scats and hair samples were genetically analysed and all data was analysed in the framework of site occupancy modelling. Detection dogs yielded the highest number of species (lynx) detections (n = 52/camera traps n = 26/ hair traps n = 3), of grid cells with lynx detections (n = 21/camera traps n = 16/ hair traps n = 2), and of unique detections (n = 14/camera traps n = 8/hair traps n = 0), meaning grid cells in which only one method detected lynx. The detection probability based on the average effort per method and grid cell was 0.75 for detection dogs for 10 transect kilometres, and 0.57 for camera and 0.11 for hair traps for four weeks of deployment each. The theoretical minimum effort required for detecting lynx in a grid cell (given presence) with a high probability (P = 0.95) was 22 transect kilometres for detection dogs and 14 and 98 weeks for camera and hair traps, respectively. The number of individual detections was the same for detection dogs and camera traps (n = 26) and was n = 2 for hair traps. Camera traps detected the largest number of different lynx individuals (n = 16/detection dogs n = 11/hair traps n = 2). The gross time effort (including passive periods of time) in days per species and individual detection was lowest for the simultaneous deployment of the two detection dog teams (species detection: detection dogs 0.7/camera traps 1.5/hair traps 13.7; individual detection: detection dogs 1.4/camera traps 1.5/hair traps 20.5). The use of detection dogs cost less (779 €) per species detection than camera traps (842 €) but more (1,558 €) than camera traps (842 €) per individual detection. Due to the low number of detections, the use of hair traps was relatively expensive (3,506 € and 5,260 € per species and individual detection, respectively). For the wolf study one scat detection dog team at a time was deployed in seven side valleys of the Magadino Plains (averagely 9.8 km2 each) in south-eastern Switzerland in a total of four two-week-periods in 2018 and 2019. Over the course of time three different dogs and two dog handlers were deployed. In each side valley, four camera traps and one audio recorder were also installed. Scat samples were genetically analysed and all data was analysed in the framework of site occupancy modelling. Camera traps yielded the highest number of wolf detections (probable wolf pictures were included, n = 104/detection dogs n = 27/ audio recorders n = 6). Dogs detected wolf scats in the highest number of side valleys (n = 5/camera traps n = 4/audio recorders n =1) and produced the highest number of unique detections (n = 2/camera traps n = 1/audio recorders n = 0). The detection probability based on the average effort per method and side valley was 0.71 for detection dogs for 10 transect kilometres, 0.99 for camera traps for averagely 20 weeks, and 0.26 for audio recorders for averagely 15 weeks of deployment. The theoretical minimum effort required for detecting wolves in a side valley (given presence) with a high probability was 24 transect kilometres for detection dogs and 13 and 148 weeks for camera traps and audio recorders, respectively. Detection dogs produced 12 individual detections of four different wolf individuals. Data from camera traps and audio recorders could not be used for individualising wolves, but camera traps detected a maximum number of five and audio recorders of two wolves simultaneously. The gross time effort (including passive periods of time) in days per species detection was lower for camera traps (1.4) than for detection dogs (2.2) and audio recorders (20.7). Costs per species detection were lower for camera traps (217 €) than for detection dogs (917 €) and audio recorders (3,228 €). Detection dogs were the only method that produced individual detections, with an effort of 5.0 days and 2,064 € per detection. The studies show that detection dogs are a suitable method for lynx and wolf monitoring, especially when high levels of data verifiability, genetic profiles (lynx and wolf) and individual detections (wolf) are required and when only short time periods are available for surveys. Camera traps were successfully used for the monitoring of both species (only species detection for wolf), while hair traps (lynx) and audio recorders (wolf) had low success rates and were relatively time-consuming and cost-intensive.