This interview with Prof. Bruce Berndt indicates the percentage of correct results from his notebooks to be greater than 99.7%. (See also this longer writeup.)

Between 1903 and 1914, before Ramanujan went to Cambridge, he compiled
3,542 theorems in the notebooks. I have gone through every entry in
the notebooks. If a result has already been proved in the literature,
then I just wrote the entry down and said that proofs can be found in
this literature and so on.There are a number of misprints. I did not
count the number of serious mistakes but it is an extremely small
number - maybe five or ten out of over 3,000 results. Considering that
Ramanujan did not have any rigorous training, it is really amazing
that he made so few mistakes.

Bruce Berndt, Ramanujan's Notebooks, parts I--V.

*side question:* The Ramanujan–Petersson conjecture for Maass forms is still open.