Objective: The Brief Illness Perception Questionnaire (B-IPQ) is a frequently used measure of illness perception (IP). The aim of this study was to explore the psychometric properties of the questionnaire when used in elderly people with knee pain. Method: Based on data from the Frederiksberg Cohort on elderly people reporting knee pain (N = 836), the psychometric properties of the eight B-IPQ items (1 ‘Consequences’, 2 ‘Timeline’, 3 ‘Personal control’, 4 ‘Treatment control’, 5 ‘Identity’, 6 ‘Concern’, 7 ‘Coherence’, and 8 ‘Emotions’) were analysed using Rasch analysis to establish whether the questionnaire provides reliable and valid measures of IP. Results: Threshold disordering was found on the 1–10 rating scale in all items. When rescaling to a 0–2 rating scale, disordering was resolved in six items. Item goodness-of-fit analyses revealed that two items displayed underfit misfit and four items overfit misfit; hence, the B-IPQ does not present unidimensionality. The person separation index indicated that items separate respondents into only two IP levels. Finally, floor and ceiling effects were found, indicating that the B-IPQ may not fully describe the extent of IP in this population. Conclusion: The 1–10 rating scale used in the Danish B-IPQ targeting knee pain is suboptimal, while a 0–2 scale improves the psychometric properties of the scale. The B-IPQ does not present unidimensionality and the use of a sum score is therefore not recommended when assessing IP. The B-IPQ may not cover the extent of IP in elderly people with knee pain.