Less than a month after Michael Brown was shot and killed by a law-enforcement officer in Ferguson, Mo., the municipal police department issued 50 wearable video cameras to its officers so they could record encounters with the public. Since then, at least a dozen other U.S. cities—including Miami Beach, Fla., and Flagstaff, Ariz.—have announced similar plans. The response is commendable, but police chiefs should proceed cautiously.

Proponents argue that the small, tamper-proof cameras will lead to fewer violent encounters between police officers and citizens because everyone knows that their speech and actions can be retrieved later. The evidence supporting such a conclusion is preliminary, however. Blindly adopting the technology without a carefully thought out policy and without training on how and when cameras should be used could make matters worse.

“What if video doesn't get recorded during a critical incident because officers are not trained, or they don't understand how to maintain the equipment?” asks Michael D. White, a professor of criminology at Arizona State University, who recently assessed body-worn cameras for the U.S. Department of Justice. A community that has learned not to trust civic authorities might suspect a cover-up. And the chances of this kind of mistake are fairly high: in one survey, nearly one third of public safety agencies using body-worn cameras did not have a written policy governing when or under what circumstances they should be activated.

Even when video images are available, they are not always conclusive. For instance, after watching surveillance recordings of a 2012 arrest in Denver, in which the head of a handcuffed woman was slammed into a wall, the police chief concluded the use of force had been appropriate. But the city's independent monitor found it excessive. Still, more evidence in most cases, even if it is not always conclusive, may turn out to be helpful.

Tantalizing hints that camera use could minimize clashes exist in the five small field trials that have been published so far. Although several of them were subject to biases because conditions were not well controlled, the tests nonetheless suggested that, overall, body-camera use decreased the number of times officers resorted to force, as well as the number of times citizens complained about police behavior.

More rigorous study is needed. Patrol areas chosen to pilot the devices should be carefully compared with similar neighborhoods where officers do not wear cameras. These comparisons should be done before and after deployment to establish a proper baseline against which to measure the results. And video recording should be compared with other efforts, such as community outreach programs or officer training to de-escalate tense situations, to see which tactics prove more effective at reducing clashes.

Research should also address important civil-liberty questions. Could the images be used to monitor or otherwise entrap law-abiding citizens? Within police ranks, some officers worry that an unsympathetic supervisor might troll videos for minor infractions to torpedo an officer's career. Who has access to the videos? Will eyewitnesses be less willing to speak forthrightly if their conversations are recorded?

The National Institute of Justice, the research and development arm of the DOJ, is funding two larger camera studies in Las Vegas and Los Angeles that should explore a few of these issues. Results are expected starting in late 2015.

Chances are that the movement to adopt body-worn cameras is unstoppable. The American Civil Liberties Union, a traditional opponent of surveillance, has cautiously embraced the technology. This momentum makes the urgent need for clear rules and training guidelines all the more apparent. Towns and cities that are planning to use the cameras should ensure that the community has an ongoing say in those plans, as well as a mechanism to resolve disputes when videos are subject to contradictory interpretations.

Finally, the DOJ, which will probably end up subsidizing the purchase of many of these cameras, should buy devices only for police forces that participate in larger research efforts and share the results with the wider public. This way we can all see what is going on.