Two sets of criteria have been proposed to discriminate spondylarthropathies (SA) from other rheumatic diseases. To evaluate their performance, we conducted a cross-sectional study in the patients observed during one week in 28 French Departments of Rheumatology by 91 staff-teaching physicians. The physicians had to apply these criteria to all their patients and had to classify them as definite SA, definite other rheumatic disease or possible SA. The analysis performed on the 2,088 patients with a definite diagnosis (124 SA and 1,964 controls) showed the following results: (table; see text) Of the 140 patients with possible SA, 37 fulfilled both sets of criteria, 22 the ESSG criteria alone and 12 the Amor criteria alone. These data suggest that a) the overall performance of these two sets of criteria is similar; b) this performance is better in the group of patients with a definite diagnosis; c) the patients without a definite diagnosis require a longer follow-up to assess the clinical relevance of these two sets of criteria.