The Insall-Salvati ratio for the assessment of patellar height was the first widely used index to be relatively independent of knee flexion. Despite the subsequent publication of other indices, it remains the most widely used index. The Insall-Salvati ratio nevertheless has drawbacks, one of which has received little attention: the ratio lacks sensitivity to patellar morphology. Unusual patellar shapes can lead to misleading values of the Install-Salvati ratio. In particular, the authors have found a significant number of inaccurate readings for patellar height in patients whose patella exhibits a long distal (nonarticulating) facet. A modified Insall-Salvati ratio is described which, when applied in conjunction with the traditional ratio, significantly reduces these errors. The cutoff point between normal and alta for this modified Insall-Salvati ratio is 2. One half of the cases of patella alta in this study were missed by the traditional ratio, yet identified by the modified index.