We find a meta-data set (715 families, up to 1,124 sib pairs) for bipolar illness to have a strong signal in a 10 cM region around D18S40, and excess paternal sharing on the q arm near marker D18S64. We describe a method for meta-analysis of microsatellite marker data using affected sib-pair (ASP) methodology. Inherent difficulties in such analysis include heterogeneity of allele frequencies and protocol design, measurement errors in genotyping, and map construction. Using identity-by-descent (IBD) allele sharing as the dependent variable, a logistic regression to test for heterogeneity finds only mild heterogeneity, and a limited parent-of-origin effect.