We develop several formal models for comorbidity between multifactorial disorders. Based on the work of D. N. Klein and L. P. Riso, the models include (i) alternate forms, where the two disorders have the same underlying continuum of liability; (ii) random multiformity, in which affection status on one disorder abruptly increases risk for the second; (iii) extreme multiformity, where only extreme cases have an abruptly increased risk for the second disorder; (iv) three independent disorders, in which excess comorbid cases are due to a separate, third disorder; (v) correlated liabilities, where the risk factors for the two disorders correlate; and (vi) direct causal models, where the liability for one disorder is a cause of the other disorder. These models are used to make quantitative predictions about the relative proportions of pairs of relatives who are classified according to whether each relative has neither disorder, disorder A but not B, disorder B but not A, or both A and B. For illustration, we analyze data on major depression (MD) and generalized anxiety disorder (GAD) assessed in adult female MZ and DZ twins, which enable estimation of the relative impact of genetic and environmental factors. Several models are rejected--that comorbid cases are due to chance; multiformity of GAD; a third independent disorder; and GAD being a cause of MD. Of the models that fit the data, correlated liabilities, MD causes GAD, and reciprocal causation seem best. MD appears to be a source of liability for GAD. Possible extensions to the models are discussed.