The probability of carrying the gene for Huntington's disease can in many cases be estimated in the children of affected persons by identifying a specific DNA marker that is genetically linked to the gene. We studied 47 persons at 50 percent risk of inheriting Huntington's disease who requested a presymptomatic or prenatal genetic-linkage test between September 1986 and January 1988. The participants were given pre-test counseling and psychological and neurologic evaluations. Nineteen persons later voluntarily withdrew from the protocol, including one who would have been excluded anyway, and one person was from a family that was too small to allow testing. Three D4S10 restriction-fragment-length polymorphisms produced by the HindIII, EcoRI, and Bg/I enzymes were used for all tests, and the probability that a subject was a Huntington's disease carrier was calculated. The accuracy of the test was compromised by a 4 percent recombination frequency between D4S10 and the Huntington's disease gene. Fifteen presymptomatic tests and one prenatal test were completed. Four yielded positive results, seven yielded negative results, and five were uninformative; seven persons are awaiting test results. All participants with positive tests experienced intermittent depression, but none required hospitalization, and no suicide threats were reported. Five participants received a diagnosis of Huntington's disease on the basis of the neurologic assessment. We conclude that some persons in the early stages of Huntington's disease may seek presymptomatic testing rather than neurologic evaluations.