Ever since it became clear in the 1990s what a valuable crime-solving tool DNA matching is, we have supported the expansion of the state’s DNA databank — which now includes all felonies and most misdemeanors. Prosecutors last week called on the Legislature to expand it again, not just for those convicted of these crimes but for those arrested, and we are in favor of that, too — with certain caveats.

DNA matching has revolutionized crime investigations, providing strong scientific evidence that can not only help convict the guilty but exonerate the innocent. In fact, the most famous cases have been used to show that a significant number of people on death row didn’t commit the crime they were accused of — a good argument against the death penalty and its finality.

Most people and organizations have come to accept taking samples from convicted criminals. But taking them from arrestees has raised objections from defense attorneys, who say it ignores the presumption of innocence, and the New York Civil Liberties Union, which says DNA information isn’t as reliable as some people think it is.

The presumption of innocence argument can be answered by the fact that fingerprints are now routinely taken upon arrest. Although fingerprints are used primarily for identification purposes and to show whether someone has a criminal history, they can also be used as an investigatory tool, matching prints that were found at a crime scene with those in the FBI’s databank. While physically there is a difference between the two types of evidence, philosphically there is none.

The NYCLU objection has some merit: DNA samples are not fool-proof. They can be contaminated, for instance, and prosecutors tend to use statistics in court that exaggerate the odds against a match — by comparing the sample statistically to the entire population rather than to the number in the databank. This overstates the certainty of the match and understates the chances of a false one.

National agreement is needed on how DNA statistics are to be used in court, and also on how DNA matching is to be used. It should never be enough for a conviction by itself, but rather a tool to justify looking at a suspect further, and to corroborate other evidence found at the scene and gathered from witnesses. The larger the database becomes, the greater the likelihood that problems will arise. But the potential benefits, for both crime fighting and justice, are worth it.