Do you factor the risk of violent death into your planning when choosing which countries to visit?

If you were to judge solely by the number of travel advisories issued by the U.S. State Department, Mexico would be the world’s most danger-prone country for U.S. overseas travelers. Indeed, far more Americans have been killed in Mexico — 598 between 2009 and 2016 — than in any other country. Afghanistan, where 84 Americans were killed during the same period, is a distant second.

But that doesn’t mean Mexico is the most dangerous country, because the raw data don’t account for the fact that vastly more Americans visit Mexico than Afghanistan. What’s needed is an analysis that puts deaths in the context of travel volume.

With that in mind, data.world looked at the numbers of Americans killed in other countries compared to the numbers of American visitors to those same countries. Ranking countries according to the number of deaths per 100,000 visitors yields the following most-dangerous-10 list, from most to least risky:

1. Pakistan
2. Thailand
3. Philippines
4. Haiti
5. Honduras
6. Nigeria
7. Belize
8. Guyana
9. Egypt
10. Mexico

In a perfect world, the number of State Department warnings might be expected to roughly mirror the above ranking. If, for example, Americans are 421 percent more likely to be killed in Pakistan than in Mexico, as the data show, State should be issuing a correspondingly higher number of Pakistan advisories than Mexico advisories. In fact, between 2009 and 2017, 28 Mexico warnings were issued, versus 25 for Pakistan.

Related: U.K. Joins U.S. Laptop Ban as Skepticism Mounts

Indeed, the connection between real danger and State Department advisories is spotty overall. In some cases, State gets it right. But elsewhere, they fail to provide appropriate warnings. For Belize, Guyana, and Guatemala, for example, no warnings were issued between 2009 and 2017, in spite of the relatively high levels of danger in those countries. And many other countries get more warnings than are justified by the actual death risk.

The takeaway from the data.world study is that travelers shouldn’t rely on State Department advisories when assessing a country’s safety quotient. They’re as likely to over- or understate the risk as they are to get it right. And for the State Department, the disconnect between warnings and actual risk should prompt them to reassess how they choose which countries to warn citizens about, and how often they issue those warnings.

After 20 years working in the travel industry, and almost that long writing about it, Tim Winship knows a thing or two about travel. Follow him on Twitter @twinship.

This article first appeared on SmarterTravel.com, where Tim is Editor-at-Large.