This study was designed to assess the analytical sensitivity and rate of agreement between commercial methods and reagents, among the most used in Italy for the detection of autoantibodies to extractable nuclear antigens (ENA). Sixty-eight serum samples from patients with clinically diagnosed systemic rheumatic diseases were aliquoted and distributed to 4 hospital laboratories; three ELISA (Elias, Shield, Inova) and 1 immunoblot method (Euroimmun) were used. Overall agreement between the test reagents, for each anti-ENA specificity, was 69.1% for Ro/SSA, 83.3% for La/SSB, 70.6% for RNP, 73.5% for Sm, 91.1% for Jo1, and 82.3% for Scl70. Lack of specificity (i.e., false positive reactions) was the most important cause of low concordance. When the data were analysed according to the clinical diagnosis, total agreement and specificity improved. However, a significant difference in terms of sensitivity was observed in the SLE group (30 sera) for RNP (positivity ranged from 20% to 43%) and for Sm (from 7% to 37%), and in the Sjögren's syndrome group (13 sera) for anti-La/SSB (from 8% to 38%). Comparable data were obtained for anti-Ro/SSA (from 70% to 77%) both in the SLE and the Sjögren's syndrome group. Sensitivity of all 4 reagents was good in detecting anti-Scl70 autoantibodies in the 8 patients with diffuse systemic sclerosis, as well as anti-Jo1 autoantibody in the 5 polymyositis patients, with a 100% and a 95% agreement, respectively. These data suggest the need of a better standardization of commercial reagents and analytical procedures, and the opportunity that every laboratory should perform anti-ENA determination by at least two different methods, since none of the methods tested was completely reliable in detecting all anti-ENA autoantibody specificities.