The stool antigen test for Helicobacter pylori is a noninvasive immunoassay to diagnose active infection with Helicobacter pylori in human fecal samples. Its performance in children and teenagers has been tested in some developed countries, showing a sensitivity and specificity above 90%, however, its accuracy in developing countries and in children under 6 years is not well established. To determine the accuracy of the test for diagnosing Helicobacter pylori infection in children and teenagers, we evaluated 133 patients (4 months to 17 years old). The gold standard was a positive culture or positive histology and rapid urease test. The test was done according to the manufacturer's instructions. However, modifications were introduced for better reproducibility. Samples were analyzed twice and results are expressed as optical density (OD) determined spectrophotometrically at 450 nm. HpSA was considered positive at OD > or =0.160 and negative at OD <0.140. One hundred twenty-seven of the 133 (95.5%) patients were included. There were no infected infants. The test showed a 94.6% sensitivity (95% CI: 90.6-98.5) and a 96.5% specificity (95% CI: 93.3-99.7). At ages 2 to 6 years the specificity was 96.4% (95% CI: 85.1-99.2) and the sensitivity was 80.0% (95% CI: 64.8-89.7), at ages 6 to 10 years the sensitivity was 100.0% and the specificity 95.7%, and above 10 years the sensitivity and specificity were 100.0%. We conclude that the test is efficient in adolescents and children, however there is a need for further studies with a greater number of patients for evaluation of its accuracy in infants.