Liabilities to complex disorders, discussed in the accompanying paper, present difficulties in measurement related to the arbitrariness of diagnostic threshold definitions and problems with discrimination between trait values, especially within the 'normal' individuals. The inability to quantitatively estimate the risk for a disorder, such as substance use disorders (SUD), is an obstacle for studying etiological (e.g. genetic) mechanisms and developing efficient prevention and treatment measures. Based on the concept of common liability to SUD, this paper delineates an application of the longitudinal family/high-risk design and item response theory to the development of a continuous index of liability. The method has been tested in both simulation study and empirical data. The approach described affords the opportunity to quantitatively estimate the risk for SUD at an early age and before any drug exposure. This method is also applicable to measuring liabilities to other complex disorders, especially those with relatively late onset.