Errors have been detected in the computer program used to analyze the data in the report entitled "Multiple Threshold Transmission of Affective Disorders" by Baron et al (Archives 1981;38:79-84). The statement that the data fit the single-major-locus hypothesis is therefore incorrect and, accordingly, conclusions based on that statement must be withdrawn. The errors detected do not invalidate the raw data. Although we lack statistical support, the raw data in our view may be interpreted as consistent with a genetic spectrum (or threshold) hypothesis whereby bipolar illness is more deviant genetically than unipolar disorder. The precise mode of transmission, however, cannot be specified.