I here investigate whether there is any version of the principle of charity both strong enough to conflict with an error-theoretic version of nominalism about abstract objects (EN), and supported by the considerations adduced in favour of interpretive charity in the literature. I argue that in order to be strong enough, the principle, which I call “(Charity)”, would have to read, “For all expressions e, an acceptable interpretation must make true a sufficiently high ratio of accepted sentences containing e”. I next consider arguments based on (i) Davidson's intuitive cases for interpretive charity, (ii) the reliability of perceptual beliefs, and (iii) the reliability of “non-abstractive inference modes”, and conclude that none support (Charity). I then propose a diagnosis of the view that there must be some universal principle of charity ruling out (EN). Finally, I present a reason to think (Charity) is false, namely, that it seems to exclude the possibility of such disagreements as that between nominalists and realists.