I have just read this article here: http://www.e-n.org.uk/6419-Unapologetic-Christianity.htm. Bart Ehrman, the subject of the article, has estimated "400,000 variations" among New Testament manuscripts, which looks like a big number.
From my own work, I can report that there are approximately 5000 words in the Greek text of just one manuscript, Codex Bezae, that display "variation" by not being spelled in the standard way. There are also about 5000 Greek NT manuscripts altogether, so assuming Bezae is indicative we can predict that there may be about twenty-five million variant spellings in the entire corpus. (This estimation is far from exact for various reasons I won't go into, but it may well be about right.)
This is far more variation than even Ehrman estimates. But the point that needs making (and the article linked does make it) is that this sort of variation is basically irrelevant - it does next to nothing to change the meaning of the texts. Simply identifying variation, to whatever degree, does not prove the texts unreliable. What matters is variation that actually makes a difference to the meaning of the text - and there is not nearly so much of that, and when it does crop up there are generally reliable ways of working out what the original probably said.