We investigated the efficiency of discarded drink containers compared with other methods widely used in small mammal studies (live trapping and diet of generalist predators). We collected 225 beverage containers (bottles and cans) from 44 sampling places, and 376 small mammals of seven small mammal species were identified. Species accumulation curves emphasized significant differences between methods, with higher species density detected by genet scats, intermediate by live trapping, and lower species density by discarded drink containers. The frequency of small mammal guilds in bottles and cans was significantly biased, shrews being oversampled and rodents undersampled, with a reversed pattern in genet scats. Our results suggested that the efficiency of discarded containers was limited by several factors: spatial issues concerning small sampling area (few square meters) and aggregation, temporal issues regarding long-lasting (and undetermined) effects in the field, trapping issues related to multiple capture power (capturing one or more individuals simultaneously), container size, selectivity (shrew-biased), and low detectability of some common species.