Twenty-four families with adult onset polycystic kidney disease were typed for markers flanking the PKD1 locus on chromosome 16. The aggregated results gave a significant lod score in favour of linkage to PKD1. Within this group of families two showed unusual features: recombinations, including double recombinations, and, in one family, an unexpectedly high proportion of affected people. We consider the evidence that in these families the disease might result from a mutation at a different locus, PKD2, not linked to PKD1. We suggest that a useful test is to compare the relative numbers of meioses apparently non-recombinant and doubly recombinant for markers flanking the normal disease locus, ignoring meioses recombinant for only a single marker. Using this test, neither our two families nor the data published so far on other families provide compelling evidence for the existence of a second locus for adult polycystic kidney disease. For genetic counselling in families too small to give internal evidence for or against linkage, the extra uncertainty can be handled by using a higher recombination rate.