During 1988 we recorded all Colles' fractures (n = 530) in a prospective study in the city of Bergen. According to defined criteria, the fractures were treated by immobilisation in plaster cast with or without reduction; reduction and immobilisation by a Hoffman external fixation device; or open reduction and internal fixation. Included in the present study were 26 patients with secondary displacement after immobilisation in a plaster cast (8% of the reduced fractures). The fractures were remanipulated and splinted in a new cast for four more weeks. The patients were followed up five years after the fracture, and underwent subjective, radiographic, and functional evaluation. The anatomical end result was significantly improved compared with the initial deformity. The dorsal angulation improved significantly compared with the "slipped' position, whereas the radial length did not. The functional end result was excellent or good in 20 patients (77%). The total movement in all directions was correlated negatively with ulna plus, and the grip strength correlated negatively with the degree of osteoarthrosis. There were no other significant linear relationships between anatomical and functional variables. Total pronation and supination correlated with the initial radial length and dorsal angulation, and the total movement in all directions correlated with the initial radial length.