Measuring foliar area or mass directly is destructive, and precludes long-term, repeated observations of individual trees as they suffer or recover from foliar damage. Instead, foliage cover indices are often used as a proxy for foliar mass. Patterns of fluctuations in foliage cover indices can be used to infer qualitative changes in canopy health. However, foliage cover is not necessarily linearly related to foliar area or mass, and this may confound the detection of significant foliar damage, and comparisons of herbivore browse impacts between individual trees, tree species or sites. I derived a mechanistic model to quantify the relationship between foliar area or mass and foliage cover measured as the proportion of sky occluded by leaves. This one parameter model is close to linear for single-tiered trees, but increasingly non-linear for multi-tiered trees. I compared the non-linear model to a linear model using foliage cover data from an artificial defoliation experiment on two single-tiered, sub-canopy species and from simulated photographic images of single- and multi-tiered canopies. The non-linear model had lower errors than the linear model, and errors did not increase with foliage density (leaf area per unit area), variation (of leaf sizes within and between canopies) or leaf geometry. The non-linear model can be easily parameterized from relatively low-cost observations of foliage cover, independently of empirical measurements of foliar area or mass, and is applicable to a wide range of tree species. It should therefore help managers quantify how changes in foliage cover due to natural fluctuations or foliar damage affect foliar area and mass, and can be used to quantify parameters for models of browse impacts in mixed forest.