Some sources of information are good and some are not so reliable. For instance
Complete Peerage is reckoned one of the very best but even after its latest volume
of Corrections and Additions, there are errors to be found in it -
Phillips' excellent further 'Corrigenda et Addenda'. It is inevitable:
people will work from the best documents they can find but years later someone
uncovers another document containing clearer information. This means that there
are few good old books in genealogy; some new books will have more and better
The basic standard of genealogical information is to find reliable documents that have survived from the time of the people concerned. I have not made it my business to find such documents myself but I have tried to find information from people that have done that.
My core standard has been Complete Peerage, but that is of little use for ancestors and relations of the last five centuries who were not peers. Fortunately various people have themselves researched their own families and done this by finding contemporary documents, much as my brother Martin has done this for the Powyses and Lybbes. So, by accumulating copies of their researches, I have built up a good quality set of sources. Faris' "Plantagenet Ancestry" and Bodine and Spalding's "The Ancestors of Dorothea Poyntz" have both been particularly good at bridging the gap between Tudor and Stuart folks and mediaeval lines.
Finally I have used Burke. Burke did little research himself; he asked his clients to produce their own family histories which he then put into sonorous prose and published. Some families did a good job of this and some did a poor job, the same as Martin found with the earlier Powys and Lybbe genealogies. So anything from Burke has to be taken circumspectly: it may be good, or it may have errors; I have generally not included anything that looked suspicious (have a look at my notes on the Goodwins of Winchendon to see where my hackles were roused).
One source that has to be treated with extreme care is anything from the internet. I have excluded anything that quoted no sources. But some people, particularly from the medieval genealogy news group, may be found regularly as a source; this is because they have done excellent research themselves and either quote, or promise to publish later, their sources.
By and large I have tried to give a source reference to everyone in my database. I have only been able to do this since my genealogy program was updated a few years back to permit this. Added to this change was my acquisition of Complete Peerage: I have been going through every peer and their ancestors in my database and documenting them from CP; this is took ages, in March 2000 I had reckoned I had done about two thousand people, but in May 2002 when I finished it I found that it was just over six thousand people that have a reference in Complete peerage - no wonder it took so long! In 2003 I went through all the Harleian editions of the Heraldic Visitations, recently out on CDROM. In 2004 I completed going through Keats-Rohan's master Domesday duo. I have done a scrutiny of The Scots Peerage but this was in conjunction with discovering the very well connected Cockburn family of Langton, Berwicks. I have accumulated some 18th and 19th century county books which are useful, some of them well researched but with a danger of accepting a pedigree in the manner of Burke; but they all give information not to be found elsewhere. And I have yet to return to the books I started with so this means that there are still a few people with no reference against them: so be careful of these people.
Novermber 2012: I have just found from the Lost Cousins November newsletter that the US 'Board for Certification of Genealogists has published a Genealogical Proof Standard which is thoroughly to be recommended; see.
In other words, for each entry have a look at the reference given and use that to judge the quality of the information given in my database. And, before you do this, take a good look at Leo van de Pas' appraisal of all the main published sources.