The size and shape of 10 native forest patches were related to tree and shrub species richness in the pre coastal mountain range in Osorno province, southern Chile. Four regression models were adjusted (lineal, logarithm, exponential, and power regression) between patch size (area = x) and tree and shrub species richness (y). Patch shape was quantified through three indexes (Patton diversity index, compactness index and fractal dimension) which were correlated with the tree and shrub richness. Results allowed to conclude that (1) species richness tends to increase significantly with patch size; this relationship was explained by the four study models. Nevertheless, the best adjustment was achieved through simple lineal regression, (R-2 = 65.1%; y = 5.9 + 0.15x); (2) patch shapes were complex in general being alike to fractal objects, and (3) area was the most consistent factor explaining tree and shrub richness (r(xyz) = 0.699); shape did not have a significant effect on variation of tree and shrub species richness.