Svrha ovog rada bila je digitalizirati strukturne karte i kartu paleoreljefa zapadnog dijela Savske depresije i napraviti 3D model koji bi se kasnije mogao koristiti u regionalnim istraživanjima ležišta ugljikovodika. Korišteno je šest karata koje su konstruirane po plohama elektrokarotažnog repera (Tg) i makera (Rs7, Rs5, Z', Rφ i α'). Navedene karte je konstruirala prof. emer. dr. sc. Josipa Velić za potrebe svoje disertacije. Digitalizacija navedenih strukturnih karta i karte paleoreljefa napravljene su u računalnom programu ArcMap 10.1, a trodimenzionalni model podzemlja u računalnom programu PetrelTM 2014. Modelirano područje se najvećim dijelom nalazi na području grada Zagreba te Zagrebačke i Sisačko – Moslavačke županije. Modelirano područje obuhvaća 960 km2. Konstruirano je 78 rasjednih ploha kojima su određeni glavni smjerovi pružanja i njihov razmještaj u prostoru. Izračunato je i statističko odstupanje izrađenog modela koje iznosi u prosjeku 0,6348 % što potvrđuje veliku točnost i preciznost prilikom izrade modela u odnosu na izvorne karte.Abstract: The purpose of this theis was to digitalize the structural maps and the paleorelief map of the western part of Sava Depression and make a 3D model that could later be used in regional exploration of hydrocarbon reservoirs. Six maps were constructed in total. One by pre-Neogen border (Tg) and five on E-log markers (Rs7, Rs5, Z ', Rφ and α'). The maps were originally constructed by PhD Josipa Velić, Professor Emeritus for her doctoral thesis made in the 1980'. The digitization of the structural maps and the paleorelief map were made in the computer program ArcMap 10.1 and the three-dimensional model of the subsurface was made in the computer program PetrelTM 2014. Research area is mostly situated around the city of Velika Gorica along with Zagreb and Sisak - Moslavina county, which roughly coveres 960 km2. The surfaces of 78 faults were designed and defined. Calculated statistical deviation of the model was an average of 0.6348%, which confirms high accuracy and precision when comapred to the original data