Until the advent of mutation analysis it was impossible to make a certain diagnosis of Huntington's disease (HD) in the absence of a positive family history, and sporadic cases of possible HD presented a substantial diagnostic dilemma. We have looked for the characteristic expanded trinucleotide (CAG) repeat sequence in the HD gene in 44 patients with probable or possible HD who did not have similarly affected relatives. We used two methods, the traditional widely used method, which estimates both the CAG repeat and the flanking CCG repeat and gives the CAG length by subtraction, and the more precise CAG method, which estimates the repeat length directly. With the CAG method, the HD mutation was detected in 25 (89%) of 28 patients with the typical clinical features of HD and 5 (31%) of 16 in whom the diagnosis was more doubtful. The CAG-CCG method gave results in the borderline abnormal range of repeats for 13 of the 33 patients eventually shown to have an unequivocal repeat expansion by the CAG method. Most of these patients had late onset of symptoms. There was evidence of expansion of an intermediate-length paternal allele in 1 patient and of non-paternity in another. The identification of the mutation causing HD means that it is now possible to confirm or exclude the diagnosis with confidence, even in the absence of a family history, by analysis of DNA from a blood sample. The precise method of measuring the CAG repeat, which is technically more difficult than the traditional method, may be needed to clarify results in a substantial proportion of such patients.