A computational method to determine cross-section requirements quantitatively is described and applied to the Tokamak Fusion Test Reactor (TFTR). To provide a rational basis for the priorities assigned to new cross-section measurements or evaluations, this method includes 1. quantitative estimates of the uncertainty of currently available data 2. the sensitivity of important nuclear design parameters to selected cross sections 3. the accuracy desired in predicting nuclear design parameters. Perturbation theory is used to combine estimated cross-section uncertainties with calculated sensitivities to determine the variance of any nuclear design parameter of interest. The paper extends the theory for cross-section sensitivity and uncertainty analysis and gives formulas for convenient upper-limit estimates for the variance of integral design parameters due to estimated cross-section uncertainties. The application to the TFTR activation analysis predicts an upper limit for the uncertainty of the calculated personnel dose rate from activated reactor components of∼45% due to all estimated cross-section errors. Since this upper limit is within the accuracy requirement of ≤50% for the calculated maximum allowable personnel dose rate, it is concluded that all nuclear data used for the TFTR activation analysis are adequate in this application.