ObjectiveTo assess the reproducibility of a questionnaire used in a surveillance system for risk factors of non-communicable diseases (SIVFRENT), based on continuous surveys through telephone interviews. MethodsThe same questionnaire was administered telephonically to a cross-section of 586 individuals aged between 18 and 64 years representative of the Autonomous Community of Madrid (Spain) on 2 occasions with an interval of 13 to 32 days. Twenty-seven variables related to perceived health status, anthropometry, physical activity, smoking, alcohol consumption, preventive practices, injuries and road safety were analyzed. To assess response stability, absolute agreement, unweighted and weighted kappa values, and the intraclass correlation coefficient (ICC) were used in nominal, ordinal and quantitative categorical variables, respectively. ResultsThe response rate to the second interview involving 461 people was 78.7%. The comparison of prevalences and averages from both interviews revealed that in 19 out of the 27 indicators analyzed, the relative deviation was under 5% and exceeded 20% in only 2 indicators. Absolute agreement was greater than 85% in 13 of the 20 categorical variables. In 10 variables, the kappa and ICC coefficients were greater than 0.8, in 13 they ranged from 0.6 to 0.8, in 3 from 0.4 to 0.6 and only 1 was below 0.4. A uniform pattern of sociodemographic variables associated with disagreement was not observed. ConclusionsThe results suggest that the telephone questionnaire used in the surveillance system on risk factors associated with behavior and preventive practices (SIVFRENT) is reproducible in a wide sample representative of the adult population of the Autonomous Community of Madrid.