INTRODUCTION: There is little research on identifying modifiable risk factors that predict future interference of pain with daily activity in people with joint pain, and the estimation of the corresponding population attributable risk (PAR). The present study therefore investigated modifiable predictors of pain interference and estimated maximum potential gain from intervention in adults with joint pain. METHODS: A population-based cohort aged ≥50 years was recruited from eight general practices in North Staffordshire, UK. Participants (n = 1878) had joint pain at baseline lasting ≥3 months and indicated no pain interference. Adjusted associations of self-reported, potentially modifiable prognostic factors (body mass index, anxiety/depressive symptoms, widespread pain, inadequate joint pain control, physical inactivity, sleep problems, smoking and alcohol intake) with onset of pain interference 3 years later were estimated via Poisson regression, and corresponding PAR estimates were obtained. RESULTS: Inadequate joint-specific pain control, insomnia and infrequent walking were found to be independently significantly associated with the onset of pain interference after 3 years, with associated PARs of 6.3% (95% confidence interval -0.3, 12.4), 7.6% (-0.4, 15.0) and 8.0% (0.1, 15.2), respectively, with only the PAR for infrequent walking deemed statistically significant. The PAR associated with insomnia, infrequent walking and inadequate control of joint pain simultaneously was 20.3% (8.6, 30.4). CONCLUSIONS: There is potential to reduce moderately the onset of pain interference from joint pain in the over-50s if clinical and public health interventions targeted pain management and insomnia, and promoted an active lifestyle. However, most of the onset of significant pain interference in the over-50s, would not be prevented, even assuming that these factors could be eliminated.