A lethal defect-wear model of mortality is presented which rationalizes the assumption of independent risks when death may be due to more than a single condition. Under this model, it is shown how competing risk theory and standard categorical data methods may be merged in a unified approach to the analysis of multiple-cause mortality data. The methodology is used to analyze linkages among diseases in the mortality data and evaluate the implication of the elimination of patterns of morbid states for multiple-cause mortality data from deaths occurring in 1969 in North Carolina.