The third set of records that hovers between primary and secondary sources will be the most controversial. Unlike visitation records and mug books, family Bible records are accepted by most hereditary organizations in their applications. In general, you need the title page of the Bible showing the year and place of publication and then the records themselves. So most researchers would put this type of records in the primary column rather than in the grey zone between primary and secondary sources.
However, in my view, the records are just like the visitation records and mug book records. They are often recollections. You need to see how many different handwriting styles there are. How many generations are given? Does the handwriting change with each generation? Again, these records are accurate the closer the records are in time and relationship to the person recording these events. That being said, I have used such records if in a single nuclear family such as the Pinkhams. I also annotated such a family Bible record set that progresses over several generations: Heard-Hayes Family Bible: An Annotated Guide, New Hampshire Genealogical Record (2006):15-26. Part of the article concerns itself with what's left out of the records. If a marriage or birth is overlooked is that positive proof that it didn't happen? Of course not.
My other problem with family Bible records is access. If a library or society owns the Bible, then it can be seen by any researcher. However, if the Bible is still held by a private person, it is harder to verify the information held within. For instance, I have family Bible records for my James Martin and Prudence Wickwire family, but only someone's transcription of them. I've never seen the actual Bible and it is held by a distant cousin. If I footnote to it, I set up future researchers for a possible wild-goose chase. [For now the footnote says: Family Bible Record (in the possession of descendant James Everett Martin (Jr.) in Elgin, NB)]. So, theses records can be boons, especially in the dreadful time period for genealogical records of 1750-1850 in the United States.