Passively working devices (with no operator input) that register bat calls in real time are very important in conservation and environmental risk assessment, but data on their performance and limits under field conditions are mostly missing. We characterized the recording variability among three batcorders placed in proximate vicinity (ca. 10 m apart) to each other at 157 sites in Austria (central Europe). We found this variability perplexingly high, both for bat activity and species richness estimates. Specifically, the ratio of the highest to the lowest total sequence length (all species combined) was over fivefold in 23%, and over tenfold in 8% of the sites. In only 17% of the sites, we found the same number of species for all three devices — in most sites it varied between one and five species. The maximum call ranges of the recorded bat species affected the recording variability between the devices only for short ranges (5 m) but showed similar or relatively low variability for longer ranges. There was significantly less recording variability in sites with no woody vegetation present than in sites with open to dense vegetation structure. The results clearly indicate that the common practice of deploying only one device per site and night very likely leads to several of the resident bat species being missed and produces unreliable activity estimates.