The validity and reliability under Swedish conditions of a translated and slightly modified version of the Stanford Health Assessment Questionnaire (HAQ), referred to here as the ADL questionnaire, was studied. Sixty-four patients with definite/classical rheumatoid arthritis (RA) participated in the major part of the investigation. In addition, inter-observer reliability was studied in the testing of 15 other patients with RA. The questionnaire was filled in by the patients twice (ADL Tests 1 and 2) with a one-week interval between. A physiotherapist or occupational therapist also assessed each of the patients on a sample of ADL functions (ADL Test 3). Joint mobility, grip-strength, pain, Ritchie index and ESR were likewise checked. Results indicated inter-observer reliability to be high for the ADL (r(S) = 0.98), for joint mobility (r(S) = 0.86), and for the Ritchie index (r(S) = 0.83). The test-retest reliability for the ADL questionnaire which the patients filled in (Tests 1 and 2) was high r(S) = 0.91. Results of the ADL questionnaires the patients completed were found to correlate fairly closely with the observations of the therapists, r(S) = 0.71. The validity of the scoring system was found to be sufficient, using Ward's cluster analysis for comparing the original HAQ scores with scores on all the questions included in the questionnaire. Thus, the translated and somewhat modified version of the ADL questionnaire studied here appears to possess a high degree of reliability and validity in assessing patients with RA.