We evaluated the analytical performance of a fully-automated electrochemiluminescence "sandwich" immunoassay method for the N-terminal fragment of the pro-peptide of brain natriuretic peptide (BNP). We then compared the diagnostic accuracy of this method in discriminating between normal subjects and patients with cardiomyopathy with that found with two previously described immunoradiometric assay methods for the assay of atrial natriuretic peptide (ANP) and BNP. We studied 193 consecutive patients (mean age 64.4 +/- 12.3 years, range 20-89 years, including 56 women and 137 men) with chronic cardiomyopathy and a group of 85 healthy subjects (mean age 52.3 +/- 12.0 years, 42 women and 43 men, range 20-79 years). N-terminal fragment of proBNP1-76 (NT-proBNP) was measured with a fully-automated "sandwich" electrochemiluminescence method using an Elecsys 2010 analyzer, while ANP and BNP were measured with immunoradiometric assay methods. The low detection limit of the NT-proBNP assay was 4.2 pg/ml (0.50 pmol/l), while the functional sensitivity was 22 pg/ml (2.60 pmol/l) with a working range (imprecision profile < or = 10% coefficient of variation) extended up to about 30 000 pg/ml (3540 pmol/l). Healthy women (64.3 +/- 41.6 pg/ml, 7.59 +/- 4.91 pmol/l) showed significantly higher values than men (46.9 +/- 30.9 pg/ml, 5.53 +/- 3.64 pmol/l, p = 0.0118). Moreover, age and sex were significantly and independently related to the NT-proBNP values in healthy subjects, as assessed by a multiple linear regression analysis (R = 0.389, F-value = 7.316, P-value = 0.0012). As expected, the NT-proBNP values of patients with cardiomyopathy were significantly higher than those of normal subjects and progressively increased with the severity of heart failure. The respective diagnostic accuracy of the ANP, BNP and NT-proBNP assays in discriminating between the group of normal subjects and that of patients with cardiomyopathy was tested by the response operating characteristic curve analysis. Our data indicated that the NT-proBNP assay is significantly better than either of the ANP or BNP immunoradiometric assays in discriminating affected patients from healthy subjects, especially when only patients with mild disease severity (New York Heart Association class I and II) are considered.