An index of kyphosis was estimated with a surveyor's flexicurve. A further index measured wedging in vertebral bodies in lateral chest radiographs by expressing the sum of vertical posterior heights of the lower six thoracic vertebral bodies and discs as a percentage of the corresponding anterior heights. Cross-sectional data showed that both indices increased with increasing age in each sex. Longitudinal studies confirmed that the increases were significant but varied widely in size. Although cross-sectional data suggested greater increase in kyphosis with increasing age in women compared with men, longitudinal measurements over five years did not show this. The index of wedging explained 42% and 48% of the variation in the index of kyphosis in men and in women, respectively. Additional effects due to age, detected in women only, added a further 4% to the explained variation, possibly from ageing of soft tissues. The relationship between the indices, and the additional effect of age in women, persisted in five year survivors.