Objective. Clinical studies of systemic lupus erythematosus (SLE) often necessitate assessment of previous disease activity, which is sometimes only possible by retrospective evaluation of information from the medical record. Using the Systemic Lupus Activity Measure (SLAM), the present study compared an assessment of disease activity obtained by chart abstraction with a direct clinical assessment. Methods. A chart SLAM was obtained using information abstracted from the clinic notes of 46 patients with SLE who were being followed up at a referral center. Chart SLAM scores were compared with SLAM scores derived from an independent assessor's direct clinical evaluation. Results. Direct SLAM and chart SLAM scores were correlated (r = 0.67 by Spearman's rank correlation, P = 0.0001). Chart review scores for disease activity tended to be lower, but to an inconsistent degree, reflecting both under- and overestimation of disease activity. The overall misclassification rate on a 3-category scale (mild, moderate, or severe activity) was 59%. Conclusion. Although chart SLAM scores were highly correlated with direct SLAM scores, misclassification of disease activity by chart review cannot be rectified. Thus, in research, medical record review should not be used as a proxy for the direct SLAM, nor should it be considered comparable with clinical assessments in the analysis of disease activity.