OBJECTIVES: To compare the accuracy of a diagnosis of pulmonary restriction made using forced vital capacity (FVC) less than the lower limit of normal (LLN) with the criterion standard diagnosis made using total lung capacity (TLC) less than the LLN in an elderly population. DESIGN: Retrospective analysis. SETTING: A teaching hospital. PARTICIPANTS: Five hundred sixty-four ambulatory and acute care hospital patients aged 65 to 96 underwent complete pulmonary function evaluation. MEASUREMENTS: Sensitivity, specificity, positive and negative predictive values (PPV, NPV) of diagnosis of pulmonary restriction defined as FVC less than the LLN were calculated in the overall sample and after stratification according to bronchial obstruction. Expected PPV and NPV at different background prevalence of true pulmonary restriction (5% and 15%) were calculated using the Bayes theorem. RESULTS: Low sensitivity (0.32) and high specificity (0.95) were found, with an area under the receiver operating characteristic curve (AUC) of 0.89. In participants without bronchial obstruction, specificity was even higher, although sensitivity decreased to 0.28 (AUC=0.92). The PPV was good (0.81), whereas with a low to moderate a priori probability (prevalence from 5% to 15%) the NPV was fair (≥0.89). CONCLUSION: A reduction in FVC below LLN cannot reliably identify true pulmonary restriction in elderly people, confirming previous findings in the adult population. Normal FVC, instead, can effectively exclude pulmonary restriction regardless of the presence of bronchial obstruction when the a priori probability is low or moderately high. [ABSTRACT FROM AUTHOR]