Carcass counts along roads can notably underestimate the number of wildlife casualties due to several sources of bias, including observers’ inability to detect carcasses during the surveys. A wide range of road-type and carcass-related factors can affect the detection efficiency by the observers, particularly when looking for small-sized species, such as amphibians, small mammals and reptiles. To assess the relative importance of factors such as road type (in terms of width and traffic volume), observer, carcass type and its condition as well as the location of the carcass, we carried out searcher-efficiency trials on two paved two-lane roads located in central Portugal, namely on a(i) national road, with wide lanes and shoulders (daytime traffic: ~22.8 vehicles/hour); and on a(ii)secondary road, with narrow lanes and without shoulders (daytime traffic: ~2.7 vehicles/hour). Each road was divided into 12sectionscomprising 500m in length (spaced at least 250m apart) along which carcasses of passerines were randomly placed (7-10 carcasses per section). In addition, bat carcasses were also randomly placed along the National road (3-5 carcasses per section).The trials were then completed by four independent, similar experienced observers driving a vehicle at 30-40km/hour that recorded the trial carcasses detected. All observers were accompanied (on the passenger seat) by the researcher who previously placed the carcasses, to check for their presence, current status (e.g. intact, flattened, only feathers) and position on the road (e.g. left lane, center line, right shoulder, verge). Our preliminary results show that bat carcasses are far more difficult to detect (detection rate: 0.28; 95% CI: 0.18 - 0.41) than passerines (detection rate: 0.69; 95% CI: 0.62 - 0.76). Carcass position and status seems to significantly affect the detection rates of passerines at the National road, but not in the Secondary road (except if carcasses fall on road verges). In both types of roads, the observers seem to detect carcasses of passerines equally well. The same did not happen with bat carcasses, with different observers showing considerably different detection rates. On average, almost half of the carcasses (43.6%) were removed before the arrival of the first observer (i.e., ~3 hours after its placement on the asphalt). Carcass disappearance resulting from traffic or scavengers can, therefore, negatively impact the robustness of the searcher-efficiency trials and, ultimately, the accuracy of the carcass detection rates obtained. We provide suggestions to minimize this issue as well as to improve the design of carcass detection trials in future roadkill monitoring programmes.