Hi Tony, My impression is that the information is culled from telephone company records, and perhaps from other sources (e.g., directories-for-sale of the sort used for charity/commercial solicitations). I suppose there must be some legal proviso that allows this information to be sold or otherwise made available on the internet -- probably the same proviso that formerly made printed telephone directories a possibility. In fact, I suspect some of the information on the websites may actually have been derived from optical character recognition (OCE) applied to scans of old printed directories because some of the errors I've noted seem to be OCE errors. Some of the information also seems to have been assembled by automated inferencing on primary sources (e.g., people with the same surname living in close proximity to one another may be assumed to be family members). In the olden days, one could request to have their phone number be "unlisted" and I note that my dad doesn't come up ! in the "familytreenow" listings, probably because he always kept his number unlisted. Ancestry.com has for years carried a number of "U.S. Public Records Index" volumes in their database, from which as often as not one is able to glean not only addresses and phone numbers, but actual birthdates as well! I generally use this information as supplementary/confirmatory evidence for relationships that I've already (at least) tentatively established, or for finding new leads for a trail that has otherwise "gone cold." I confess that I have also used the information from time to time (and with little luck) to attempt establishing contact with potential participants in my DNA surname project, or to attempt placing NEW project members in family trees ahead of the curve. This sort of information was also extremely valuable in enabling me to establish contact with one or two somewhat distant branches of my own family tree. I'm generally not averse to its existence but am aware of the potential for it being used for mischief. As to reliability, the main limitations seem to be, 1) inferential errors, 2) OCE errors, and, 3) that the information generally appears to be somewhat dated -- at least 5 to 10 years old. -Mark Bunch (#6223)