Abstract Karst is developed by the dissolution of soluble rock caused by an active fluid flow. It significantly increases the porosity and permeability compared to the matrix system. Active flowing along the top, bottom and edges of the lenses near the coastline leads to the formation of pores of meters scale because of the dissolution of the rocks. All this must be considered when constructing a geological model for further effective development of the field. A new approach in karst extraction and modeling is implemented using factor analysis. The first stage consists of a selection of large karst geobodies using seismic data. Inversion seismic volume (an inversion for acoustic impedance) is used for the extraction of geobodies. In addition, a variety of seismic volume attributes is used to extract karst geobodies, and then a resampling of these seismic attributes is done. After that synthetic logs for all wells with non-matrix properties on the attribute parameters were obtained. Geobody extraction followed by its export to the properties of the geological model is carried out. The second stage includes the extraction of karst using well data, such as core, well logs, mud losses intervals and CHCD (closed hole circulation drilling). Then factor analysis is used to examine the link between the values of different variables and determine the best correlation between the parameters and groups. Based on these data karst flags on wells are generated. As a result, generated discrete karst log is scaled up and distributed throughout the geological model using a 3D trend of geobodies property in Petrel software. The percentage of karst in geobody is based on well data. The size and distribution of karst are based on field analogues and published materials on outcrops analogues. Porosity logs within karst intervals are created and distributed in the 3D model using a stochastic algorithm. Own methodology of karst model construction is presented, especially new approach in karst extraction and modeling is applied using multivariate statistical analysis - factor analysis. Factor analysis is multivariate method applied to study the link between the values of variables. It is assumed that the known variables depend on fewer unknown variables and random errors. The implemented mathematical apparatus is confirmed with real data on the well.