The identification of a DNA restriction fragment length polymorphism closely linked to Huntington's disease on the short arm of chromosome 4 has for the first time allowed presymptomatic prediction to be undertaken in first-degree relatives at risk. The late and variable onset of this dominantly inherited disorder makes such prediction a powerful and potentially valuable aid in genetic counselling, but in the absence of effective therapy there are serious ethical reservations concerning such a predictive test. The new developments have stimulated an active and informative debate among professionals and family members on whether and how predictive tests should be used. Guidelines have emerged which should be useful not only for Huntington's disease, but for other serious late-onset neurogenetic disorders. Meanwhile, studies in Wales and elsewhere have not only confirmed the original linkage but have excluded multi-locus heterogeneity as a significant problem. Genetic prediction for the individual at risk remains critically dependent on a suitable family structure, present in only a minority of families in Wales. A more feasible alternative for most families is prenatal exclusion, which can allow risk prediction for a pregnancy without altering the situation for the person at risk. This approach has already been applied in Wales; the experience gained will be useful in full prediction, which is currently being introduced.