Schizophrenia shows considerable clinical variation, but the relationship between clinical variables and degree of genetic loading for schizophrenia is unclear. We investigated this by analyzing published data from the adoption study of Kety et al. (1994) in Denmark. We sought to determine which clinical variables in proband adoptees with chronic schizophrenia predicted risk of schizophrenia in their biological relatives, using logistic regression analysis. We found that risk of chronic schizophrenia in relatives was predicted by the presence of pervasive negative symptoms (odds ratio [OR] = 9.44, 95% confidence interval [CI] = 1.98-45.01) and absence of pervasive positive symptoms (OR = 0.09, 95% CI = 0.01-0.78) in probands. Pervasive negative symptoms were defined by the presence of all of the symptoms: social withdrawal, autistic behavior, poverty of thought/speech, and flat affect. Pervasive positive symptoms were defined by the presence of all of the symptoms: suspiciousness/ideas of reference, delusions, auditory hallucinations, and other hallucinations. These clinical variables may be useful for refining phenotypic definitions of schizophrenia in molecular genetic studies.