The presence of a prevalent vertebral deformity increases the risk of a future vertebral fracture. The aim of this study was to determine whether certain characteristics of the prevalent deformity, including its shape and location in the spine, influenced this effect. The 3100 men and 3500 women who took part in this analysis were recruited from population registers for participation in the European Prospective Osteoporosis Study (EPOS). Subjects had lateral thoracic and lumbar spine x-rays at baseline, and again after a mean interval of 3.8 years. Prevalent morphometric vertebral deformities on the baseline film were identified by the McCloskey-Kanis method. Incident fractures were defined as vertebrae that also satisfied the McCloskey-Kanis criterion for prevalent deformities on the follow-up film, and in addition had at least one height (anterior, mid, or posterior) which had reduced by at least 20% between films. Poisson regression was used to assess the association between various characteristics of the prevalent deformity and the risk of an incident vertebral fracture, with generalised estimating equations used to allow for the fact that each subject contributed several vertebrae to the analysis. The risk of an incident fracture increased with the number of prevalent deformities: relative risk (RR) for one prevalent deformity 3.2 (95% confidence interval (CI); 2.1, 4.8); 9.8 (95% CI;6.1, 15.8) for 2; and 23.3 (95% CI;15.3, 35.4) for 3 or more. Relative risks differed significantly according to the shape of the prevalent deformity, ranging from 5.9 (95% CI; 4.1, 8.6) if the anterior and mid heights were reduced to 1.6 (95% CI;0.8, 3.2) if the posterior and mid heights were reduced. Risks varied also according to the severity of the deformity. There were fivefold differences in relative risk of incident fracture depending on the location of the prevalent deformity within the spine. Compared to vertebrae in subjects with no deformities at baseline, the relative risk of an incident fracture within three vertebrae of a prevalent deformity was greater (7.7 (95% CI;5.6, 10.5)) than the risk in more distant vertebrae (4.0 (95% CI;2.6, 6.0)). In summary, the risk of a subsequent vertebral fracture in individuals with preexisting deformities is importantly influenced by the characteristics of these deformities.