Evidence from epidemiological studies and segregation analysis suggests oligo- or polygenic inheritance in schizophrenia. Since model independent methods are thought to be most appropriate for linkage analysis in complex disorders, we performed a genome-wide autosomal screen in 71 families from Germany and Israel containing 86 independent affected sib-pairs with parental genotype information for statistical analysis strictly identity by descent. We genotyped 305 individuals with 463 markers at an average distance of approximately 10 cM genome-wide, and 1-2 cM in candidate regions (5q, 6p, q, 8p, 10p, 18p, 22q). The highest multipoint LOD scores (ASPEX) were obtained on 6p (D6S260, LOD = 2.0; D6S274, LOD = 2.2, MHC region, LOD = 2.15) and on 10p (D10S1714, LOD = 2.1), followed by 5q (D5S2066, LOD = 1.36), 6q (D6S271, LOD = 1.12; D6S1613, LOD = 1.11), 1q (D1S2675, LOD = 1.04), and 18p (broad disease model: D18S1116, LOD = 1.0). One hundred and thirty-three additional family members were available for some of the families (extended families) and were genotyped for these regions. GENEHUNTER produced a maximum NPL of 3.3 (P = 0.001) for the MHC region and NPL of 3.13 (P = 0.0015) for the region on 10p. There is support for these regions by independent groups. In genome-wide TDT analysis (sTDT, implemented in ASPEX), no marker passed the significance level of 0.0001 given by multiple testing, but nominal significance values for D10S211 (P = 0.03) and for GOLF (P = 0.0032) support further the linkage results on 10p and 18p. Our survey of 22 chromosomes identified candidate regions which should be useful to screen for schizophrenia susceptibility genes.