The objective was to compare four different scoring methods for vertebral deformities: the semiquantitative Kleerekoper score and three quantitative scores (according to Minne, Melton and Raymakers) in patients with rheumatoid arthritis (RA). Lateral radiographs of the thoracic and lumbar vertebral column were scored in 52 RA patients treated with corticosteroids (Cs) and 52 RA patients not Cs treated (matched for age and gender). Clinically manifest vertebral fractures were defined as a vertebral deformity leading to the prescription of therapy: bedrest and/or analgesies. The number of Cs-treated RA patients with vertebral deformities was only increased according to the Kleerekoper score (relative risk 2.31; 95% confidence interval 1.36-3.90). The total number of vertebral deformities was increased in Cs-treated RA patients according to the method of Kleerekoper. Minne and Raymakers, but not according to Melton. The Spinal Deformity Index was increased in the Cs-treated RA patients according to Minne (P < 0.05), but not according to Raymakers. In both patient groups, the number of patients with clinically manifest vertebral fractures was much lower than that of vertebral deformities, e.g. in the Cs-treated patients 13% vs 35.79% (depending on the scoring method used). All eight patients with clinically manifest vertebral fractures had severe height loss (25% or more) in at least one vertebra. Vertebral deformities seem to occur more often in Cs-treated patients than in those not Cs-treated. However, differences exist in the number of patients with vertebral deformities and in the total number of vertebral deformities between different scoring methods. This emphasizes the need for a gold standard or consensus for defining vertebral deformities, in which, in our opinion, a Spinal Deformity Index has to be included. Clinically manifest vertebral fractures are predominantly characterized by severe vertebral deformities on X-rays.