In April 2009, police easily arrested Alonzo J. King, Jr. in Wicomico County, Md. After King pointed a shotgun at a group of people, one of them told the police who did it, and King readily admitted his guilt. He was originally charged with felony assault and ended up pleading guilty to a misdemeanor.

This seemingly insignificant criminal case is now before the Supreme Court, with arguments later this month. That's because of what the authorities did next. When King was arrested, police took a cotton swab of skin cells from inside his cheek for DNA testing. They did not need his DNA to link him to the shotgun incident. Instead, the police entered King's DNA profile into both the Maryland DNA database and the FBI's national database, CODIS. King's profile, like all those in the database, was then automatically compared every week to evidence from all unsolved crimes. And, in fact, King's DNA matched DNA from an unsolved sexual assault case, for which he was later convicted and sentenced to life in prison.

If Maryland had required King to submit his DNA once he was convicted, then there would be no Supreme Court challenge. So far courts have all upheld DNA collection from felons, reasoning that convicts forfeit some of the rights of ordinary citizens. Maryland v. King is about something new: More than one-half of the 50 states (including Maryland) and the federal government authorize compulsory collection of DNA from people who have been arrested. But the Supreme Court has never held that if police have probable cause to arrest, they can also search a suspect for evidence of past or future crimes. Maryland's justification for this unprecedented expansion of police power? Bigger is better. Add arrestee profiles to the database, and more crimes will be solved.

Wrong. Research shows that bigger is only better if DNA databases grow in the right way: by entering more samples from crime scenes, not samples from arrestees. DNA databases already include 10 million-plus known offender profiles. But a database with every offender in the nation cannot solve a crime if no physical evidence was collected or tested. And police collect far too few such samples.

Police do routinely collect physical evidence in cases of homicide and in most cases of rape. But evidence is not collected from eight out of 10 crime scenes for other serious offenses, like burglary, robbery and aggravated assault. Forget what you see on the proliferation of CSI spinoffs. Many jurisdictions do not even have dedicated and trained crime scene investigators.

States like California, which vastly expanded DNA databanks to include arrestees, do not generate dramatically more matches between offenders and crime scenes than do states with much smaller databases, like New York or Illinois. That is because New York and Illinois, despite the smaller numbers of offenders in their databases, enter crime scene samples at rates comparable to California. Indeed, from 2010 to 2012, California halved the average number of offender profiles uploaded per month, but kept the number of samples from crime scenes constant. The result was an increase in database hits. The same dynamic played out in the United Kingdom. The lesson is clear: The police solve more crimes not by taking DNA from suspects who have never been convicted, but by collecting more evidence at crime scenes.

Even worse, taking DNA from a lot of arrestees slows the testing in active criminal investigations. After all, 12 million or more people are arrested each year. (According to one study, by age 23, nearly one-third of Americans have been arrested for an offense, not including minor traffic violations.) Backlogs created by arrestee DNA sampling means that rape kits and samples from convicted offenders sit in storage or go untested. This hurts innocent suspects, like Cody Davis, who was convicted in Florida because of a delay in testing evidence that later cleared him. And although you likely have a right to have your DNA expunged if your arrest was mistaken, in some states it is up to you to petition the court, and needless to say there is no right to counsel or easy online way to carry out such a request.

Putting DNA from arrestees into databanks also exposes more innocent people to the risk of false accusation or conviction. Interpretation of DNA evidence from known offenders is straightforward, but crime scene samples often require subjective judgments that may lead to errors. What is more, cross-contamination and accidental sample switches have occurred in labs across the country. In one case in Nevada, a man spent four years in jail because an analyst accidentally switched his sample. In three more cases, erroneous DNA testing led to wrongful convictions that were overturned by subsequent DNA tests. A 2009 National Academy of Science report criticized the current lack of quality control in the forensic testing system. But improvement seems less likely if crime labs are inundated with DNA from arrestees. The FBI has also opposed confidential access for researchers who could independently assess government assertions about the accuracy of DNA databases.

In addition, arrestee testing exacerbates the racial disparities in DNA databases. Because African-Americans and Hispanics make up a disproportionate share of convicts, they are overrepresented in databases. Racial disparities in arrest rates, particularly for minor crimes like drug possession, can be even starker. Allowing states to bank DNA of arrestees will mean including disproportionate amounts of genetic information from African-Americans and Hispanics as compared to other groups.

King's case may seem like a poster-child for DNA database expansion, because his sample closed a serious unsolved case. But there is more to the story. King had at least six prior convictions at the time of his sentencing for the shotgun offense. A state law limited to taking DNA from convicted offenders could have authorized his testing. The same is true of dozens of other arrestee DNA "success stories."

That suggests that the line between arrest and conviction is the right place to locate the constitutional limit on the government's unending appetite for collecting our DNA. If the Supreme Court approves Maryland's law, then DNA samples will be routinely taken from people upon arrest just like fingerprints. Unlike fingerprints, however, the DNA samples will be used not to identify people, but to forever implicate them as one of the "usual suspects." Also unlike fingerprints, DNA samples convey a lot of information, like your sex, what you look like, whether you are adopted, or how old you might be. Some states have even decided, without judicial approval or new laws, to use DNA databases to find not only people in the database, but also their family members. And as technology evolves, who knows what is next? What we know now is that no good evidence shows that taking from DNA from arrestees helps to solve crimes in any meaningful way. We also know the cost to liberty and privacy. That's the tradeoff the Supreme Court should weigh.

Brandon L. Garrett is a professor of law at the University of Virginia School of Law who studies criminal procedure, civil rights, and wrongful convictions. His new book, "Convicting the Innocent: Where Criminal Prosecutions Go Wrong," was published by Harvard University Press. An interactive website about the book, called "Getting It Right," will soon be hosted on the Innocence Project website.

Erin Murphy is a professor of law at New York University School of Law. They wrote a friend-of-the-court brief in Maryland v. King.