According to Bureau of Justice Statistics numbers here, as of 2002 21% of state prison inmates were in on drug related charges.
From the same source, also for 2002, 25% of jail inmates were in on drug related charges. 11% were simple possession, the balance were distribution.
From the second source, 55 of federal prisoners were drug related as of 2001 but that was not broken down by offense.
That is still too many, but wildly exaggerated claims of easily verifiable numbers do all sides a disservice.
All of the statistics are deceptive. This is because most convictions are not made just for single offenses. Prosecutors learned years ago that they could turn the chances for conviction in their favor by filing what amounts to redundant charges. This is because juries, when offered a choice, are far more likely to choose from the list, rather than to go for the guilty and not guilty choice.
In turn, this skews the drug statistics terribly. Many people who are arrested for drugs do something during the arrest that is an easier conviction than the drug charge, such as trying to run away from the police. So even though they are purely a drug conviction, they are in prison for resisting arrest along with their drug conviction.
Importantly, such other convictions are often much easier because they have no objective evidence, only testimony.
In real terms, in the State of Arizona some years ago, a legal decision basically ordered the release of all small quantity drug possession cases in the State prisons, assuming they had no other conviction.
The State prison system released one man, declaring that no one else was eligible for release because of other offenses.