When forest soils are investigated, the identification of litter and soil layers is a key step. Mid-infrared spectroscopy seems to be promising for this due to its capability to provide fingerprint information. In this study, a large data set of mid-infrared spectra of different forest soils (L, FH and Ah horizons of coniferous, mixed and deciduous forests) was available together with general analyses data (among others organic carbon, total nitrogen and carbonate contents). This set was used for devising models for the identification of different soil layers and to follow the characteristics of different forest types in the different layers. Canonical discriminant functions were calculated using nine mid-infrared bands occurring in both organic layers and mineral soil layers. The discriminant analysis of the topsoil layers showed that aliphatic mid-infrared bands dominated the first function and yielded a clear separation of mineral and organic layers. Further, this analysis reflected the higher chemical diversity in the organic layers visible by a high scattering of the calculated data points. Two tests with two separate data sets showed a reliable and reproducible performance of the model. Separations of forest types were best possible in the organic layers, in which the deciduous forests were separated best from mixed and coniferous forests. Also for these discriminations, aliphatic bands were dominating. Advanced degradation of soil organic matter resulted in small distances of equidistant forest-type group-centers in the 5–10 cm layers of the mineral soils. Dominating molecular moieties in the discriminant functions in these layers were carbonates and aliphatic bands. The applied methodology suggests mid-infrared spectroscopy as an appropriate tool for soil layer identification, allowing for unequivocal discrimination between organic and mineral soil layers. Discriminations of forest types delivered information about dominating mid-infrared bands in the single layers and allowed conclusions about functional groups dominating these differences. ► Mid-infrared spectroscopy was able to separate organic layers and mineral soils. ► Forest types were better separated in the organic layers due to fresh litter. ► Organic layers: highest contributions to the models from aliphats, amides, alcohols. ► Mineral soil layers: highest contributions from carbonates and aliphats.