I now have data from a fourth Trio of parents/chid data and in this case the data is in both gedmatch.com and FTDNA. In this case 100% of the "Distant Relative" matched either the mother or the father. For the category of Speculative Relative I have now been able to analyze 5 cases of child matches where neither parent matched and I could find all four sets of data in gedmatch.com. Only one of those cases was real, meaning the segment of 7.87 cM in the child did not match in either parent. The other 4 cases were all "False Negatives" on the part of FTDNA where they should have found a match but did not. In this set of data FTDNA found 216 "Speculative Relatives" and 41, or about 20% did not match one of the parents. From the small sample above that would mean 20% of the 20%, or 4%, are true Identical By State(IBS), or just random matches and that 96% are solid Identical By Descent, IBD, matches. I note that most of the FTDNA non-matches, 29 of them, happened in the range of 8.50 cM and less. There are 60 matches in that range so the mismatch ratio there is about 50%. The one true mismatch that I found was at 7.87 cM by FTDNA measurement I looked at that one mismatch in some detail on gedmatch.com (I hope this table formats OK when you view it) Person Chr Start Location End Location Centimorgans (cM) SNPs Child 10 111192330 116724473 7.3 1988 Father 10 111293348 114533483 3.5 1076 Mother 10 112273709 115111806 3.8 1027 One difference is that gedmatch.com got 7.3 cM and FTDNA got 7.9 cM another is 1988 SNPs per gedmatch and 1500 per FTDNA. That may be due to gedmatch using 23andMe SNPs and FTDNA using their fewer SNP data. Note that on gedmatch.com neither the parental nor the maternal phase of the child had any match to this genotype. Thus it looks to me that if you have a FTDNA match to a person at any level and that segment also matches a parent, aunt, uncle, grandparent or cousin then you can call that segment Identical By Descent(IBD) and look for the common ancestor. This data suggests that FTDNA matching algorithm for Speculative Relatives gives 96% IBD, (good matches) and 4% IBS random matches. This is in contrast to the finding that gedmatch.com at 8 cM will find some 20 to 50% of the matches are just random IBS matches. Thus the algorithm that FTDNA uses to establish a match does eliminate more IBS matches than the straight cM and SNP count used by gedmatch.com See this for details on FTDNA matching http://www.familytreedna.com/faq/answers.aspx?id=17#608 __________________