The barren karst areas are not typically for the karst regions of the morphoclimatic system of temperate zone. Nevertheless the importance of assessing the extension and degree of compactness of bare karst is grounded by the fact that these areas represent both particular geomorphological landscape, with specific implications in the karstic processes, and diverse and fragile ecosystems. The touristic potential of barren karst is important too. The paper presents a method for automatic detection of the barren karst using Object – Based Image Analysis (OBIA) in the central sector of the Mehedinţi Mountains, an area where bare limestone have a large extent. The object-based classification offers the advantage that it takes into account both spectral and qualitative data. Thus we have used in analysis a complex set of data: multispectral SPOT images, two derived indices – Normalized Difference Rock Index (NDRI) and Normalized Difference Vegetation Index (NDVI) - geological data and the digital map of built-up areas and roads. Results were compared with very high resolutions SPOTView colour images and GPS field data. The high accuracy of the classification, over 94%, demonstrated that the method is precise enough to be used in geomorphological mapping of the karst.