This work is a critical analysis of the term "flexibility of scoliosis." The authors demonstrate that this measure includes two very different and noncorrelated phenomena: the collapse and the reducibility of the structural curve. In 228 paralytic curves, these two phenomena were quantified and correlated with different parameters of scoliosis: sex, stage of maturation, curve pattern, number of vertebrae in the curve, spinal and abdominal test values, and the scoliotic evolution rate. Reducibility can be predicted since it is an elasticity process. Collapse is more complex, but it is a prognostic factor for evolution and effectiveness of treatment. The preliminary results of a similar study of idiopathic scoliosis are the same.