As there is little information on the factors that influence fracture union following intramedullary nailing of the tibia we retrospectively investigated patient-, injury- and treatment-related factors in 161 patients with closed or grade I open fractures of the tibial diaphysis. The patients were reviewed until clinical and radiological evidence of union at a mean of 13.3 months (4 to 60). Multivariate statistical analysis using a Cox proportional hazards model showed that the risk of failure of union increased by 2.38 times for highly comminuted fractures, by 3.14 times when nail dynamisation was applied, and by 1.65 times when the locking screws failed. In fractures with no or only minimal comminution the risk of nonunion increased if the post-reduction gap was > or = 3 mm.