From what I've observed in the other answers, there are generally two opinions: either all errors are unacceptable, or some minor ones are acceptable. (Have you noticed a potential run-on sentence just now? Because "some minor errors are acceptable" is an independent clause next to an independent clause, I would've caused a run-on sentence had I not separated both clauses thereof with a comma prior to the conjunction "or." This is similar to what another user, Scott, failed to notice when he wrote, "All you can do is try, everything else is a gamble" without a conjunction, period, or anything to properly separate these two independent clauses.)
But despite me being an analytic philosopher, whose occupation literally implies truths that are contained within definitions, I'll have to agree that grammatical errors are almost inevitable from a practical standpoint; almost no books ever published are error-free, and depending on one's knowledge of the language used, more errors will appear.
I noticed that the users who were harsher on errors committed ones in their own responses. For instance, spiceyokooko, an interesting choice for a username, wrote in his or her second paragraph, "I would check each paragraph for errors until I was satisfied they were all correct" before submitting his or her answer. What I found intriguing is the missing punctuation by the et cetera abbreviation, etc., which should always end in a period; had it been a well-known acronym (as opposed to an abbreviation), then he or she would not have needed to use any periods with respect to it.
Another error occurred when spiceyokooko didn't use a period following the abbreviated versus (i.e., v.) and another when "400 page book" was written without a hyphen; the latter error was repeated four times! (Someone hasn't been proofreading!) So, what do these tell us? In a matter of 279 words, spiceyokooko made at least six grammatical errors (seven, if we count the missing Oxford or serial comma near the end of his or her response) and probably more if I were less impatient in finishing this response and more patient in reading his or hers for additional errors.
The point of the above example isn't to denigrate anyone, but it's to show that in almost a standard book page's worth of text, even someone who appears to be well-versed in English and committed to reviewing just 279 words will make several blunders, ones that are pretty easy to spot (i.e., a missing hyphen, missing periods following abbreviations, etc.). It's to say that—to answer your question, sharptooth—the average number of mistakes in a published book could be somewhere near five mistakes per page, if someone is a decent writer; yes, five mistakes per page would be exceptional. The reason why these mistakes aren't being found is that even copyeditors, who could cost between .2 and .4 cents per word or 2.5 to 5 dollars per page, only know a fraction of all the grammatical rules, and of those that copyeditors do know, their understanding of them isn't 100%. For example, I haven't met a single English teacher who properly understood the definite article, despite it being by far the most used word in the English language! How did this happen? How did every philosopher I have ever known likewise fail to understand, by far, the most used word in the English language? This is unfortunately the dirty truth of an author's job; one would have to be a hypocrite to tell you that all mistakes are unacceptable while committing mistakes himself. This is why, when I talk about grammatical correctness, I review and review what I write so that I am setting an example for others. So far, I reviewed my response five times before hitting submit, and I submit that there might be an error somewhere, but perhaps the error would be found by someone who knows a less accessible rule of grammar than the ordinary English reader.
If someone had a full understanding of grammar, there would no doubt be the occasional contradiction within grammar itself, but the best advice I can give is that if you want the safest bet of what passes as "acceptable," despite standards that change from publisher to publisher, just make no grammatical mistakes at all; stick to safe sentences that you know you can write well, avoid staying up late at night, upsetting social situations with friends or work colleagues, avoid drinking, make sure to review everything you write, and then print and review your work. Repeat this process sentence by sentence until you can recite your book with your eyes closed. While it's certainly time-expensive and emotionally expensive, it's what separates mediocre authors (sometimes they are erroneously hailed as "greats" within our generation) from truly magnificent and legendary ones (i.e., the ones that are born once a century). The alternative is to keep errors to about five or fewer per page, and, if we're being honest with ourselves, you're doing fine by authors' standards.
Thanks for reading my response. (I respond to people.)