Early on in last night’s presidential debate between Hillary Clinton and Donald Trump, when the questions turned to crime and justice, Trump spoke like a tough-on-crime politician from the 1980s. Saying that we “have a situation where we have our inner cities, African Americans, Hispanics living in hell because it’s so dangerous,” he argued that we “have to bring back law and order.” He then pushed for expanding the use of stop-and-frisk, stating that “murders are up.”

At first glance, the 2015 Uniform Crime Reports—the FBI’s assessment of crime rates nationwide, which was released yesterday—seem to back up Trump’s concerns. There were more violent crimes in 2015 than in 2014, particularly murders. Compared to 2014, 2015 experienced about 1,500 more murders (an 11 percent increase), 6,000 (or 5 percent) more forcible rapes, 4,500 (or 1 percent) more robberies, and 33,000 (or 5 percent) more aggravated assaults. (Most property crime, however, continued to fall, with burglaries down 8 percent and thefts down 2 percent.)

The picture Trump painted, however, was unnecessarily bleak. Dangerously so, in fact. The American criminal-justice system is inherently prone to overreact to the slightest bit of bad news. In Arkansas, for example, a 10 percent decline in prison populations suddenly turned into a 25 percent rise when a single murder in 2013 caused prosecutors, judges, and parole officers to act much more harshly. It is imperative that we not overreact to the new crime statistics, and there are at least four major reasons for us all to remain calm.

Crime remains at historic lows.

Despite the increases cited in yesterday’s FBI report—the rise in murders in 2015 was the largest in both absolute and percentage terms since crime started dropping in the early 1990s—the United States remains an historically safe place to live. The murder rate in 2015 is still lower than it was in 2009, and before 2009 the last time the murder rate was as low as it was last year was in 1964. Overall, 2015 had the third-lowest violent crime rate since at least 1970, and probably even before that, since our older crime stats likely understate crime much more than they do today.

Yes, crime went up in 2015. But crime remained at near historic lows in 2015, too. Both of these statements can be, and are true. Despite the rise in violent crime, we remain safer today than we have been in decades.

One year is never a “trend.”

People often talk about the “steady decline in crime” since the early 1990s, but that’s not entirely accurate. Between 1991 and 2014, the number of murders fell from almost 25,000 to slightly more than 14,000—but the number of murders rose during seven of those 23 years, including four years in a row from 2000 to 2003. Forcible rapes rose in seven years as well, robberies in three, and aggravated assaults in four. In fact, 2005 and 2006 saw two consecutive years of increases in every crime except forcible rape (which rose only in 2006). Yet even then we continued to talk about declining crime—despite two years of increases and despite the fact that levels were higher then than they are today. That’s because one and two year changes are not trends.

Now, to be fair, the magnitude of the increase this year, especially for murder, is striking. The one-year increase in murders between 2014 and 2015—1,532—is greater than the total increase in murders over the four years between 2000 and 2003 (1,006). The size of this jump should not be glossed over. But it is not a trend.

Crime rates can and do move unexpectedly. After all, the last time we saw an absolute or percentage rise in murders similar to this one was between 1989 and 1990, and just two years later, violent crimes pivoted from rising sharply to declining (somewhat) steadily. Putting too much weight on a one-year move, however large, is unwise.

We are political victims of our own successes.

One statistic we are likely to hear a lot in the days and weeks ahead—one that is uniquely deceptive—is that 2015 saw the largest percent increase in murders, rapes, and aggravated assaults since 1990. In fact, the percentage jump in murders was actually larger than that in 1990. Strictly speaking, these statements are correct, but their implications are the exact opposite of what those who invoke them are intending to say.

Because we have so much less violent crime today than in 1990, any given increase will be a bigger percent jump today than 25 years ago. If we have 100 units of something, five more is just 5 percent, but that same five-unit increase is a 10 percent jump from 50. So while the number of murders rose by 11 percent in 2015, compared to 9 percent in 1990, the total increase in murders in 2015 was about 400 less than in 1990. The percent change looks worse because we are doing so much better.

The crime rise is concentrated.

When crime rose between 1989 and 1990, it was a broad, nationwide increase.

Nearly half of the 75 cities with populations of over 250,000 saw increases in murder, robbery, and aggravated assault, and 71 cities saw at least one of those crimes increase. Furthermore, the increases were generally proportional to the cities’ levels of crime: Cities with more crime saw bigger (absolute) increases. The biggest increases in murders occurred in New York City, Houston, Chicago, Los Angeles, and Dallas, among the biggest cities in the country.

The rise in 2015 looks somewhat different. Only 30 percent of the 73 cities with populations over 250,000 saw increases in murder, robbery, and aggravated assault, and seven experienced declines in all three types of crime. Moreover, the increase was much more concentrated: a greater fraction of the increase occurred in a smaller number of cities, and those cities tended to be relatively smaller. Unlike in 1990, the cities with biggest increases in murders, for example, were Baltimore; Chicago; Houston; Washington, DC; and Milwaukee. Such a pattern makes the increases seem less the natural result of a broad, nationwide rise in crime, and more changes arising from local conditions.

There is, however, one unappreciated wrinkle with this argument that must be acknowledged. In 1990, 65 percent of the national increase in murders occurred in cities with populations over 250,000; in 2015, only 44 percent of the increase was in cities that large. So within cities, the rise in murders and other violent crimes is more concentrated today than in 1990. But the overall increase is somewhat more diffuse. While it is again too early to talk about any sort of “trend,” in the push to avoid overreacting we should not simultaneously ignore troubling indicators.

So yes, crime rose in 2015, and for murder, rape, and aggravated assault, it rose by more in percentage and absolute terms than it has since crime started to decline in the early 1990s. Yet the truth is the country remains safer than at almost any time since crime began to rise in the 1960s. And given our inclination to overreact to any sort of bad news about crime, it is important not to lose sight of that critical context.