The mass shooting in Las Vegas Sunday night that killed at least 59 people and injured hundreds more was one of the worst such massacres in American history — that’s something everyone who studies gun violence can agree on. It was also the seventh mass shooting so far this year. Or was it the 337th?

These two numbers come from competing efforts to track mass shootings in the United States. (A third lists the Vegas massacre as the 272nd mass shooting of 2017.) One effort follows the federal definition of “mass killing” of three or more people killed; another counts nonfatal injuries, too.

The differences in these counts matter less than the injuries and loss of life resulting from all types of gun violence (more than 33,000 people die from guns in the U.S. each year, a third of them in homicides). But they illustrate a key dispute over how we measure the problem, one that could have important implications for what we choose to do about it.

Each effort makes a pretty good case for why its method is best — so rather than pull out one big number on mass shootings, here’s a breakdown of the most significant of these efforts and their tallies.

The biggest reason we can’t put a clear number on the frequency of mass shootings is that nobody can agree on what a mass shooting really is. Should you tally the number of people killed? The number injured? Does the gunman count as a victim if he’s killed by police or takes his own life? Must the shooting occur in a public place, or do domestic violence cases matter too?

These are grim questions to have to answer, and research on gun violence in general is far behind where it could be. Until recently, efforts to look specifically at mass shootings were few and far between.

Congress limits research into gun violence and its causes through a 1996 law that prohibits the Centers for Disease Control and Prevention from advocating for gun control. Although the law doesn’t explicitly ban research on guns, it’s had a chilling effect in the two decades since its passage, so much so that a group of prominent public health researchers in February called on their colleagues to work to reverse it.

Law enforcement agencies collect data on mass killings, and the FBI includes this information in its annual Supplemental Homicide Reports. But, in part because the data is self-reported, a 2014 USA Today investigation found inaccuracies in more than half of cases.

A number of independent, journalistic, and crowdsourced efforts attempt to fill this gap, but they vary by methodology and by what kinds of incidents they track. That’s why we end up with such differing counts and headlines that say “One mass shooting occurs every day” countered by headlines that say “No, it doesn’t.”

So before you cite a statistic on mass shootings, it’s worth understanding where it comes from and how researchers arrive at their numbers.

Mother Jones magazine keeps what is probably the best database of mass shooting incidents like the one that occurred in Las Vegas. Their definition is narrow: “The attack must have occurred essentially in a single incident, in a public place”; they exclude “crimes of armed robbery, gang violence, or domestic violence in a home, focusing on cases in which the motive appeared to be indiscriminate mass murder”; and there must be at least three fatalities, not counting the killer.

By their count, there were six mass shootings in the U.S. this year before Sunday’s became the seventh. They count 91 total mass shootings from 1982 through 2017, or fewer than three per year.

Journalists at Mother Jones undertook an investigation of mass shootings in 2012 after the movie theater shooting in Aurora, Colorado, and keep a database of incidents up to date. They contend that indiscriminate shooting in a public place differs from other shooting incidents, even those with multiple victims. To count them together, they argue, risks inflating the numbers on a specific — but still growing and very much concerning — problem.

Their findings track with those of criminologist Grant Duwe from the Minnesota Department of Corrections, a prominent researcher on the topic. They also track fairly well with the FBI’s most recent study of active shooters.

In 2014, the FBI published a study (PDF) of “active-shooter” incidents, using a narrow definition that seemed designed to capture the type of massacre that occurred in Vegas. Like Mother Jones, they didn’t count gang- or drug-related violence, for example, or most incidents occurring in a home.

Researchers identified 160 active-shooter incidents from 2000 through 2013, or fewer than a dozen per year. But the FBI did find that active-shooter incidents were growing in frequency, from 6.4 per year in the first half of its study to 16.4 per year in the second half. They counted a total of 486 people killed (not including the gunmen) and 557 people wounded.

This was still three years before the 2016 shooting at the Pulse Nightclub in Orlando that, until Sunday, was this country’s mass shooting with the most victims: 49 were killed and 58 were wounded.

Most of the incidents in the FBI study had far fewer victims — the median number killed was two, as was the median number wounded. And just two in five met the updated 2013 federal definition of “mass killing” that set the threshold as three or more people killed in a single incident.

The gun control advocacy group Everytown for Gun Safety released a report in April examining mass shootings from 2009 through 2016. They used an older federal definition of “mass killing” as four or more people killed and found 156 mass shootings over those seven years, or nearly two dozen per year.

The larger annual number is because Everytown counted domestic violence incidents — which it said made up more than half of incidents — that the FBI did not include in its active-shooter study. They found that a quarter of all mass shooting fatalities were children.

Like Everytown, two other major efforts to track mass shootings include incidents that go beyond the FBI’s active-shooter and Mother Jones’ mass shooting definition. As a result, they count far more incidents each year.

This independent effort, started in 2013, draws from law enforcement and media reports and uses an expansive definition of “mass shooting.” Their counts include nonfatal as well as fatal shootings, so they consider any incident where four or more people are shot to be a mass shooting.

They report 990 mass shootings from 2014 through 2016, or 330 per year. Another 273 have taken place so far in 2017 by their count. Las Vegas was the 272nd — the group documented a 273rd mass shooting for the year on Monday, in which five people were shot, three fatally, overnight in Lawrence, Kansas.

This database, spawned from a Reddit community and crowd-sourced, uses a loose definition for “mass shooting” similar to the Gun Violence Archive’s but records significantly more incidents. That’s because they include among the victims the gunman and anyone else who might be shot by law enforcement responding to the scene.

They argue it makes sense to report the data this way “because, ‘shooting’ means ‘people shot.’”

They count 1,850 mass shootings from 2013 to date, and by their definition the Vegas massacre was the 337th mass shooting this year (they also count the Kansas shooting as 338th).

Whichever count you use, two things are clear: mass shootings still represent a small fraction of the gun violence that occurs in the United States each year, and mass shootings occur here far, far more often than anywhere else in the world.