Background: Increasing age at onset has been associated with worse outcome in rheumatoid arthritis, although there are few data from unselected inception cohorts. Hypothesis: Increasing age is associated with a higher risk of erosions at presentation, and this increase is not explained by age-related disease confounders. Subjects and methods: 222 subjects (median onset age 59 years) were studied from a primary-care-based register of new-onset inflammatory polyarthritis. Patients had hand and feet radiographs taken within 12 months from symptom onset. Films were scored by two readers using the Larsen score. The risk of erosions in those aged 50-69 and >= 70 years at onset was compared with the risk in those aged,50 years both before and after adjustment for possible age-related disease confounders. Result: The prevalences of erosions were 22%, 52% and 71% in those aged,50, 50-69 and >= 70 years at onset equivalent to odds ratios (ORs) (95% confidence intervals (CIs)) of 3.5 (2.2 to 5.7) and 7.4 (4.5 to 12.1), respectively, in the two older age groups. Excluding those with proximal interphalangeal (PIP) erosions alone (due to possible osteoarthritis) did not alter these findings. Adjustments for disease characteristics using logistic regression did not attenuate these findings: adjusted ORs (95% CIs) 3.6 (2.1 to 6.1) and 6.9 (3.8 to 12.2) for age groups 50-69 and >= 70 years, respectively. The influence of age was stronger than most of the disease-related variables in predicting erosions in this cohort. Conclusion: Increasing age at symptom onset is strongly associated with higher occurrence of erosions within the first year unexplained by greater disease severity.