Familial Mediterranean fever (FMF) is an autosomal recessive disorder of unknown pathogenesis, characterized by recurrent, self-limited attacks of fever with synovitis, peritonitis, or pleurisy. Using DNAs from affected Israeli families, we have recently mapped the gene causing FMF (designated MEF) to the short arm of chromosome 16, with two-point lod scores in excess of 20. In this report we consider the possibility of a second FMF susceptibility locus. Before discovering linkage to markers on chromosome 16, we had found suggestive evidence for linkage to chromosome 17q, with the following maximal two-point lod scores: D17S74 (pCMM86), Z = 2.47, (theta = 0.20); D17S40 (pLEW101), Z = 2.15 (theta = 0.15); D17S35 (CRI-pP3-1), Z = 1.78 (theta = 0.15); D17S46 (pLEW108), Z = 1.69 (theta = 0.18), D17S254, Z = 2.30 (theta = 0.20). Moreover, multipoint linkage analysis using D17S74 and D17S40 as fixed loci gave Z = 3.27 approximately 10 centimorgans (cM) telomeric to D17S40. Data with the chromosome 17 markers alone in our families suggested locus heterogeneity. Nevertheless, our families were not separable into complementary subsets showing linkage either to chromosome 16 or to chromosome 17. We also examined the possibility that the positive lod scores for chromosome 17 might reflect a secondary, modifying locus. By several measures of disease severity, families with positive lod scores for chromosome 17 loci had no worse disease than those with negative lod scores for these loci. We conclude that chromosome 17 does not encode a major FMF susceptibility gene for some of the families, nor does it encode a disease-modifying gene. Rather, it would appear that linkage to chromosome 17 is a "false positive" (type I) error. These results reemphasize the fact that a lod score of 3.0 corresponds to a posterior probability of linkage of 95%, with an attendant 1 in 20 chance of observing a false positive.