Summary1. As a fundamental unit within a landscape, habitat patches create spatial heterogeneity. On a gradient of habitat patch size, one sees patches of denser vegetation cover occurring in otherwise open habitats (e.g. bushes in the desert, forest fragments in grasslands). 2. The resulting habitat fragmentation and consequent edge effects continue to concern conservation biologists, landscape ecologists, and wildlife managers due to their role in influencing wildlife abundance and distribution on broad geographic scales. 3. Although ecological processes near habitat edges often differ from processes away from edges, an animal's suitable habitat may increase or decrease by 'positive' or 'negative' edge effects. Such effects can be measured using behavioural indicators and experiments based on the costs and benefits of foraging under diminishing returns. 4. We used live trapping, giving-up densities [amount of food left behind in experimental food patches -giving-up density (GUD)] and geographic information system (GIS) analyses to test for spatial and temporal habitat selection by the four-striped grass mouse (Rhabdomys pumilio) inhabiting a grassland containing distinct 'islands' of woody vegetation (500 m 2 of average area). We investigated how these woody islands influence the quality of the grassland in terms of activity patterns, feeding opportunities, predation risk and habitat suitability. We used R. pumilio's GUDs to explicitly map and measure habitats of varying quality using GIS. 5. Our results showed that R. pumilio perceived the wooded patches as habitats with no opportunities and yet highly risky when opportunities are provided. Furthermore, the wooded patches had a negative edge effect, their unsuitability extended for 1-3 m into the otherwise suitable habitats. 6. The combination of live-trapping and giving up densities provided an unbiased and controlled measure of a prey's perception of habitat use and foraging costs when equal opportunities are simultaneously provided. These techniques and concepts should be broadly applicable by conservationists and ecologists interested in understanding and managing edge effects. 7. This approach aims to effectively quantify edge effects and extrapolate predation risk on a landscape scale, thereby developing recommendations for managers to improve the quality or amount of edges for wildlife.