Roussouly classification: the four shapes of the sagittal spine

The Roussouly classification groups the normal variation in the sagittal shape of the lumbar spine and pelvis, standing, into four types. It is read on a lateral radiograph: the pelvis tells which shape the spine was built for, and the sacral slope tells which shape it stands in today. (Roussouly 2005)

The idea: the pelvic incidence is the signature

The pelvic incidence (PI) is anatomical. It does not change with posture or with degeneration, and PI = pelvic tilt (PT) + sacral slope (SS). (Legaye 1998)

The sacral slope orders the lordosis. In 160 volunteers, the overall lordosis, its inclination, the position of the apex and the number of lordotic vertebrae were determined by the slope. Four shapes follow from that. (Roussouly 2005)

Degeneration disguises the shape. As lordosis is lost the patient tips the pelvis back, the slope falls, and a type 4 spine can look like a type 2. The original shape is recognized only by the incidence: a very retroverted pelvis with a high incidence was certainly a type 4, and a large lordosis has to be restored. (Roussouly & Pinheiro-Franco 2011)

The current type: sacral slope and apex

The slope cuts are 35° and 45°. The apex of the lordosis is placed at the center of L5 (type 1), the base of L4 (type 2), the center of L4 (type 3) and the base of L3 or higher (type 4). (Roussouly 2005)

Type 3 with an anteverted pelvis was found in 16% of 296 asymptomatic adults: PI 48° ± 6°, PT 4° ± 3°, slope over 35°. (Laouissat 2017) The PT cut of 5° or less is the one of the school itself. (Sebaaly 2017) The MEANS study defines the anteverted variant with PT under 8°. (Shen 2024)

The theoretical type: from the incidence

Since the incidence does not change, it tells the type the spine was built for.

One in three asymptomatic volunteers has a current type different from the theoretical one. A mismatch is therefore not by itself a disease, and the spine is a continuum, not five boxes. (Shen 2024)

The two arcs of the lordosis

The horizontal through the apex splits the lordosis in two (Berthonnaud’s construction). The lower arc, from S1 to the apex, equals the sacral slope. The upper arc, from the apex to the inflection point, is nearly constant, about 20°. So the spinal lordosis, from S1 to the inflection point whatever the number of vertebrae, is the slope plus about 20° (r = 0.89 between lordosis and slope). (Chevillotte 2022)

The upper arc of the lordosis is the lower arc of the kyphosis, so lengthening it raises the low thoracic kyphosis. (Pizones 2025)

Reference values and ideal profiles

In the series that proposed those ideals, getting the type and the level of the apex right mattered most; the inflection and the distribution did not reach significance. (Pizones 2019)

Degenerative shapes

In 331 patients with a degenerative spine, the sagittal shape fell into eleven types, from the slope, pelvic tilt, apex, lordosis and thoracic kyphosis. (Sebaaly 2017)

• Classical types 1 to 4.
• Anteverted (PT 5° or less): type 3 and type 4.
• Retroverted (PT 25° or more): false type 2, false type 2 with thoracic kyphosis, false type 3. Retroverted types 1 and 4 did not appear.
• Kyphotic: lumbar kyphosis with a hypokyphotic thorax that compensates, and global kyphosis without compensation.

The path of degeneration runs from type 3 to false type 2, then to false type 2 with thoracic kyphosis and on to global kyphosis. Type 4 goes to false type 3 and follows the same path. (Sebaaly 2017)

The pelvic incidence is not always constant once the pelvis is fixed with screws, and the change differs with the type of fixation. This matters when comparing the film before and after. (Daher 2025) (Daher 2026)

How ScolioSim does it

1. Open Simulate, 2D, on the standing lateral film and use the Spinopelvic tab. Mark the femoral heads, the S1 endplate and the endplates of L1, L4 and the inflection point.
2. Choose the lordosis apex and the inflection vertebra from the lists: you read them off the film.
3. The card gives the theoretical type (from the PI) and the current type (from the SS and the apex), each measure against the range for the target type, the sacral slope and the spinal lordosis that would stand as the target type, the ideal values of that type, and the degenerative shape.

The target sacral slope is the PI minus the mean PT of the type; the target spinal lordosis is the target slope plus the mean upper arc of the type. The shape is then labeled: matched, compensated, false type, flattened, hyperlordotic or kyphotic.

Related: how a pedicle subtraction osteotomy at each lumbar level acts on the arcs is described in osteotomy planning on the radiograph; the angles themselves are explained in spinopelvic parameters.