Objective To identify factors associated with non-participation at the 12-month and 24-month follow-up phases of a prospective cohort study of injury outcomes.

Methods Associations between non-participation at follow-up phases and a range of sociodemographic, injury, health, outcome and administrative factors were examined.

Results An individual's non-participation at 12 months did not necessarily mean non-participation at 24 months. Sociodemographic factors were the most salient for non-participation, regardless of the number of follow-up phases or specific phase considered.

Conclusions Retention rates in prospective cohort studies of injury outcome may be improved by follow-up of everyone irrespective of previous non-participation, focusing resources to retain men, young adults, indigenous people and those living with people other than family members, and by ensuring that multiple alternative participant contacts are obtained. There is sufficient evidence to be concerned about potential bias given that several of the factors we, and others, have identified as associated with non-participation are also associated with various functional and disability outcomes following injury. This suggests detailed investigations are warranted into the effect non-participation may be having on the estimates for various outcomes.