New global strategies for road building require innovative tools to analyze linear patterns and their spatial distribution and to evaluate their environmental impacts. Roads not only present physical barriers to wildlife but also provide access for human and biological invasions. In tropical regions especially, forest degradation has been associated with roads built for selective logging into formerly intact forest landscapes. To quantify to what extent ecosystems are influenced by roads, it is important not only to know road length density but also their location in a landscape unit. The concept of roadless space is based on distance to the nearest road from any point. We present the computation of this distribution using the Empty-Space-Function, a general statistical mathematical tool based on stochastic geometry and random sets theory. We demonstrate the applicability of this well-defined probability function to calculate roadless space based on vector road data. In a Congo Basin case study we compared the temporal development of road networks inside different logging concessions over time. We hypothesized that roadless space decreases, even when the rate of wood volume harvest remains constant. Based on LANDSAT time series covering the last 29 years, we assessed accessible roads in relation with the river network and calculated the roadless space at different points in time. As expected, roadless space decreased continuously throughout most concessions, despite a drop in total annual harvest volume after 2008 and independent of forest certification schemes. We recommend that measures to reduce impacts of selective logging should not only be based on the extraction of timber, but should also include the total area impacted by roads. The Empty-Space-Function provides a rigorous mathematical description and a straightforward way to assess intact forest landscapes and is therefore highly applicable to road impact evaluation in conservation science.