Objective. To evaluate the usefulness of the Human Activity Profile (HAP) in predicting estimated maximal oxygen uptake (VO 2max ) in persons with arthritis and to evaluate the ability of 2 classification systems to distinguish individuals with arthritis who have poor fitness from those with average and above fitness. Methods. Forty-four subjects with arthritis completed the HAP and a submaximal treadmill test. The adjusted activity score (AAS) was derived from responses on the HAP. VO 2max was estimated from the submaximal treadmill test. The ability of the AAS and age to predict estimated VO 2max was determined with multiple regression analysis. Subjects were also assigned to a fitness category based on their AAS and estimated VO 2max , and agreement of these categories was assessed using the kappa statistic. Two classification systems were used, including one proposed by the original authors and one we proposed based on more recent normative data. Results. Sixty-six percent of the variance in estimated VO 2max could be accounted for by the AAS and age. The kappa statistic for our proposed classification system was 0.35, indicating fair agreement, whereas the kappa statistic for the original classification system was incalculable. The sensitivity of the proposed classification system to identify persons with average and above fitness was 84%, with a specificity of 50%. Conclusion. We suggest that the HAP is useful in estimating fitness level when standard exercise testing is not feasible.