From the 1990's GIS has been a part of a standard toolkit in most archaeological projects. As computational power increased, so did the spectre of potential uses of GIS in archaeology. One of those potential uses is a least cost analysis (LCA). LCA is an expremely useful tool that enables scientists to „predict“ potential routes that could've been used at a certain point in time, even if the remains of those routes have not been preserved. Besides predicting the routes that have been lost, it is also possible to reconstruct the missing routes by comparing the LCP with teh existing road network, in order to better understand them. This paper will try to offer a contribution in answering several questions regarding the Roman road network, using the exampel of wider area of historic region of southern Liburnia (today roughly coresponding to northern Dalmatia). The validity of LCA aplication in southern Liburnia has been confirmed. Reliability of the results varies depending on the microregion within the area of study. The biggest precision has been achieved in the Bukovica region. Some of the LCPs generated in this project do not satisfy the validity criteria and have been rejected. LCPs generated in the area of Ravni Kotari have a high rate of variability. Those LCPs that run paralel to the ridges in the area have a high rate of reliability, while the ones running perpendicular to the ridges have a low rate of reliability.