To assess whether the clinical knowledge of the treating surgeon had any effect on the reliability of the pain-drawing evaluation, drawings from 50 low-back pain patients were evaluated by the treating surgeon and by three colleagues who had no clinical knowledge of the patient. The evaluation was repeated after 10 days. The treating surgeons were also blinded to clinical data. The kappa value in the evaluation when the surgeon had clinical knowledge of the patient was lower (0.29 (95% Cl 0.13-0.45)) than the kappa value in the evaluations made without clinical knowledge (0.60 (Cl 0.45-0.75)). The differences observed in interobserver reliability between open and blind evaluations suggest that clinical knowledge of a patient influences the evaluation of the pain drawings.