Species inventories provide the basic information for biodiversity studies. Designing cost-effective species inventories is important in estimating the species richness or the number of additional species expected in a study area. Complete species inventories are difficult to achieve and a variety of estimation methods are available to counter the underestimation of species richness associated with incomplete sampling. A main criterion of selecting an estimation method requires deciding on a sampling unit type. The sampling unit may vary across study and estimating species richness using different sample units may produce different results. We evaluated the ability of incidence-based estimators of species richness to provide reliable estimates of species richness of temperate herpetofauna (i.e. amphibian and reptile) communities using two different types of sampling units (i.e. numbers of sampling days and the number of sampling sites, respectively). Our results showed that incidence-based estimators varied in their performance according to the two sampling units used. The incidence-based coverage estimator performed best when sampling unit was the number of sampling days whereas second-order Jackknife was the best estimator using sites as sampling unit. We draw attention on the lack of robustness of incidence-based estimators to sampling unit and recommend using species richness estimators in inventories of herpetofauna depending on the sampling unit.