We analyse the relationship between landscapescale variables (fragment area, shape, connectivity, altitude, edge contrast, forest type, etc.) and vascular plant richness in an Andean forest of northern Colombia. Using a vegetation map at a scale of 1:10000, we randomly selected 39 fragments (from a total of 700) in each of which we established a 200 m 2 plot where we counted the total number of species of vascular plants. The sample consisted of fragments of between one ha and 840 ha at altitudes of between 1800 and 2800 m. In the 60 plots (1,2 ha in total) we found 108 families, 259 genera and 556 species of vascular plants. Species richness varied between 34 and 74 per 200 m 2 plot. The analysis showed that the individual variables most closely related to species richness (according to the Pearson r coefficient) are the total area of the fragment (r = 0,43, p <0,001, n = 60), the central area of the fragment (r = 0,49, p <0,001, n = 60) and the fragment edge contrast (r = - 0,48, p <0,00, n = 60). Finally, we built multiple regression models to explain up to 32,6 % of the variation in species richness of each fragment. We discuss the usefulness of these models for the identification of small scale conservation priorities.