Selective logging prevails in tropical forests around the world, posing urgent questions about how to reconcile timber extraction with biodiversity conservation. Roads are those elements of selective logging that are most costly, most visible and they probably have the most far-reaching environmental impacts. While many studies have outlined road related threats to forest ecosystems, little is known about the persistence of logging roads in the forest landscape. This is especially important in Central Africa, where selective logging is the most important type of land use, both in terms of spatial extent and financial yield. In this thesis I analyze the temporal and spatial dynamics of logging road networks in a part of the Congo Basin and apply these findings to make suggestions for forest management. In five chapters I am approaching the subject from different angles and on different scales: In the introductory chapter, I compare the content and the orientation of scientific literature on logging roads in tropical forests. In general I identified two strains in the literature, one focusing specifically on road related impacts on forest ecosystems and the other giving technical advice in road planning, building and maintenance in order to improve efficiency and reduce impacts. A third, partially distinct direction of research is oriented on the characterization of the spatial distribution and coverage of forest road networks on larger scale to monitor forest exploitation and related degradation. The second chapter presents a methodology to identify roads in CentralAfrican forests based on remote sensing with LANDSAT images. In a time series approach, I used survival analysis to evaluate the temporal dynamics of secondary logging roads over the last 30 years and showed how road persistence differs depending on environmental variables such as geological substrates. The third chapter approaches the persistence of logging roads from a field based perspective. I carried out vegetation inventories on a chronosequence of roads abandoned between 1985 and 2015. The results showed that road tracks and edges are suitable habitats for commercial species regeneration with rapid changes in the environmental conditions occurring over time. During 30 years after abandonment about one third of the biomass lost for road building has re-captured in subsequent vegetation development. The fourth chapter analyses the extent of logging road networks in the overall forest landscape. I used the mathematically well-defined Empty Space Function as a novel way to calculate roadless space. I demonstrated how roadless space in intact forest landscapes (declared in 2000) has diminished in general but in particular in FSC-certified logging concessions. I recommend that forest management should make the preservation of large connected forest areas a top priority by effectively monitoring - and limiting - the occupation of space by roads that are accessible at the same time. The concluding chapter develops management suggestions to apply the findings. I showed that re-opening logging roads in subsequent harvests is rather the exception than the rule. Evaluating benefits, opportunities, costs and risks, I conclude that re-opening roads should be given a higher priority in forest management. Re-using logging roads can spare forests within the same area by avoiding new forest clearing in the vicinity and at a larger scale by sparing unlogged forests from new logging disturbance by intensifying operations on previously logged forests. As a vision for road management, I suggest to actively manage logging roads as transient elements in the landscape until they are reopened. Permanent access roads should only be built in the periphery of continuous forest blocks. As a perspective for further research, I discuss the trade-offs between the need of roads for development and the environmental impacts. As an example for this, I present evidence for the first major road corridor crossing the Congo Basin that is already under construction. To limit the impacts on the forest, large-scale conservation corridors have to be established, requiring supra-regional landscape planning.