One might argue "Well! It's a little late for that isn't it because you're 25,000+ people in, and perhaps you should've thought of this BEFORE you got started!" and, I suppose they'd be right. I think part of the problem is in the ancestry.com advertising: they make it sound as if there's a BIG STORY behind every ancestor (and in fact there probably is, although realistically that big story is largely lost to the winds of time), and so as you get your feet wet and "discover" so many more people, etc. you get caught up in the whole thing (which of course is the ultimate goal of the marketing department!). What newbies like me DON'T realize - at first - is that many of the "sources" or at least pointers to sources come from other newbies, and they're also not particularly adept at ensuring accuracy.
I'll admit I've found some REALLY idiotic boners, and have inadvertently created some of my own. Seeing children to people in family trees whose birth dates are long after the one or both parents are dead - or even more perplexing born before their own parents.
Of course with most of my efforts centered on Québec, there are two primary sources for information: the Tanguay and Drouin collections, respectively. Both have combed through the parish records which start in the 1620s (or when the parish was established). The ideal (and naïve) expectation is that you'll find every instance of every baptism, marriage and burial for every person who lived there; the reality is more complex, of course.
Tanguay's work is a set of seven volumes spanning 1608 to 1800 (or shortly thereafter). The first volume goes to 1700, the rest give updates for the 17th century data and then continue on for the 18th century; in reality it has more complete data up to 1750 or so and then things trickle off between 1750 and 1800. Tanguay is ordered by family name of the husband, then date of (first) marriage. From there you get the date of marriage, where it happened, then a list of children, their baptism dates, and their marriages (where known). Each husband is tagged with a roman numeral for their generation in the family line: "I" for the first appearance (i.e., the immigrant), "II" for his sons (who marry), "III" and so on. Both husbands and wives are linked to their father's entries (where available). For many of the first generation, there's also information about his parents, and where they lived (typically in France, but I have come across Vienna). So, it's a handy resource and occasionally has VERY interesting footnotes concerning their positions in society, what land they owned, how they died, and so on.
Tanguay is very convenient because it's printed, BUT it's absolutely not infallible: I've encountered SEVERAL errors, up to and including entire entries being assigned to the wrong family (specifically, two cousins with the same name being assigned to the wrong father/uncle). Errors in dates - especially pre-1700 - is common.
Drouin is (for practical purposes) a microfilm collection of as many of the available parish records as possible. It's the ultimate in "raw" data. Ancestry.com has done a so-so job annotating things, but what you get from their "Search" function is fuzzy at best: unless the volume of the parish record is for just one year, you'll get a year range that can be next to useless (e.g., "1640-1749" for a birth year). To get useful information, you have to actually look at the record (which I didn't realize for the longest time). Plus the transcriptions/identifications from the ancestry.com community is - well, amateur-based.
However, the Drouin Institute has launched their own effort to making the records accessible, the "LaFrance". It has a meager cost ($13CAD/month) but offers FAR easier access to the Drouin microfilms with a better search engine. I've been using both services side-by-side for my new effort. (You have a daily limit of 75 image views with the LaFrance, whereas if you can figure out which Drouin microfilm you're looking for, access to the images is covered with your ancestry.com subscription.) The LaFrance isn't complete: marriages to 1913, but baptisms/burials only to 1849, but they're continuing to progress in their identification. There are mistakes, but FAR fewer than you encounter in Tanguay. One thing that is INCREDIBLY HELPFUL is that LaFrance's search engine NOT ONLY handles the myriad variations on surname (and given name), but ALSO does a very good job of sort out 'dit' names (sometimes you have to provide them to have a listing show up, but when you enter in a surname, you get a list of the potential 'dit' names that exist for it. VERY handy.
So, my technique has altered:
- Look up a marriage in Drouin (through LaFrance), enter in the metadata;
- Look up the parents (if I don't already have them);
- Rinse, lather, repeat until I've determined that the spouse is/isn't a blood relative, fixing any obvious errors along the way.
 ... although there are several cases where children have been born shortly after their father has died.
 ... strangely enough I have NOT seen many cases of children born to couples who were not yet married, or were clearly born after "shotgun weddings". There are a few who were born 7-8 months after the marriage, though.