We studied 56 patients with fractures of the tibial shaft in a multicentre prospective randomised trial of three methods of external fixation. Group I was treated with a fixator which was unlocked at 4 to 6 weeks to allow free axial compression (axial dynamisation) with weight-bearing. Group II was treated with a fixator that was similarly unlocked at 4 to 6 weeks but included a small silicone spring which on weight-bearing could be compressed by up to 2 mm. this spring returns to its original length on cessation of weight-bearing thus allowing cycles of motion of up to 2 mm. Group III had a spring fixator like group II, but it was unlocked from the start to allow cyclical micromovement as soon as weight-bearing began. Fracture healing was monitored by the measurement of fracture stiffness. We defined healing as achieving a stiffness of 15 Nm per degree. The mean time was 14.1 weeks in group I, 15.9 weeks in group II, and 19.3 weeks in group III. The difference between groups was statistically significant (p = 0.004). The 95% confidence intervals for the average delay in healing with early cyclical micromovement (group III) as compared with later axial dynamisation (group I) was 1.8 to 8.7 weeks. The healing time in patients whose cyclical micromovement was delayed for 4 to 6 weeks (group II) was between these two extremes, but the differences from either of the other groups could have been due to patient selection. In the patients who completed the full trial, there were pin-track infections in over 60% of those in the cyclical micromovement groups compared with 20% in the axial dynamisation group (p = 0.03).(ABSTRACT TRUNCATED AT 250 WORDS)