Vending Machines Kill More People Than Sharks — And That's a Meaningless Comparison

The Internet’s Favorite Dangerous Fact

Somewhere in the early 2000s, a statistic escaped from a Consumer Product Safety Commission (CPSC) report and became a fixture of bar trivia, listicles, and “did you know” accounts: vending machines kill more people per year than sharks. The figure usually cited is around two deaths annually in the United States. Sharks, by comparison, kill about one American in a typical year according to the International Shark Attack File, so on US numbers the comparison is legitimately close.

The vending machine number became shorthand for irrational fear: we worry about the wrong things. We’re scared of sharks while standing next to the actual killer at the end of the office hallway. It uses accurate numbers to say something completely false.

Why It Feels Like a Fun Fact

The “vending machine vs. sharks” comparison works because it pairs two things that feel categorically different. One is a living predator, ancient and powerful, the subject of blockbuster films. The other is a beige appliance that dispenses Doritos. That cognitive dissonance makes the comparison feel like a revelation about human irrationality.

To a point, the underlying claim holds up: media coverage of shark attacks is wildly disproportionate to actual shark fatality rates, particularly in the United States. But the corrective morphed into something sloppy. The vending machine became proof that our fears are upside down, that the mundane things around us are the real killers.

The vending machine statistic does not mean vending machines are dangerous. It means one specific behavior around vending machines is dangerous, and almost nobody does that behavior accidentally.

The Exposure Problem, and Then Some

Most risk statistics suffer from a denominator problem: the raw death count is divided by the entire population, obscuring that most of the population is never meaningfully exposed. Coconut fatality rates only make sense when you count people who stand under coconut palms. Shark fatality rates need the population of ocean swimmers. Lightning rates need outdoor workers in storm-prone regions.

Vending machines are doubly misleading. There are somewhere between three and five million vending machines in the United States, depending on whose count and which decade, and Americans interact with them billions of times per year. If vending machines were generically dangerous, we would expect to see a vastly higher death toll from those billions of interactions.

We don’t, because machines sitting against walls do not spontaneously kill people. They tip over when people rock them.

The CPSC data on vending machine fatalities consistently shows the same mechanism: someone rocks or shakes the machine, typically to retrieve a stuck item or recover coins from a failed transaction, and the machine falls forward onto them. A fully stocked vending machine weighs between 400 and 900 pounds. A 700-pound object falling four to five feet onto a person’s torso or head is not survivable without immediate emergency intervention, and often not even then.

This is not a passive risk. The correct denominator is not billions of transactions but the number of times people rock vending machines, a far smaller number reflecting a behavior the machines literally carry warnings against.

The Real Math

The hard data is thinner than the fun fact suggests. In a 1995 study, the CPSC counted at least 37 deaths and 113 injuries from people rocking or tilting machines between 1978 and 1995. Spread over those 18 years, that is about two deaths and six serious injuries a year, or roughly eight incidents a year in which a machine actually landed on somebody. That study prompted the industry warning labels now on every machine, and the label campaign appears to have worked: the last reported US vending machine death was in 2008.

Nobody counts how often people rock vending machines. This site assumes about five million rocking incidents a year, roughly one per machine per year, and treats that as a site estimate rather than a measured figure. Eight tip-onto-person events among five million rocks gives a topple rate of about 1 in 600,000 per rock. Once a machine actually lands on someone, the outcome is brutal: in the CPSC record, one in four of those people died. Multiply the two and the chance of dying per rocking attempt is about 1 in 2.5 million. What keeps the death count low is not the lethality of impact but how rarely machines actually tip onto their target.

The calculation chains three things: that rocking causes a tip onto you, how often you rock, and that impact is fatal. The product is very low. But the product is not zero, and the number of people who rock vending machines out of frustration or genuine effort to retrieve their money is not zero either.

Your Actual Risk

The calculator below can model your personal vending machine risk. If you have never rocked or tilted a vending machine, your risk from this cause is functionally zero.

If you do rock vending machines, you have elevated your risk meaningfully. At the default of one rock a week for 50 weeks, the calculator returns about 1 in 50,000 per year, somewhat below the US car-crash rate of 1 in 8,000. Since the calculator is calibrated to the 1978–1995 record and no US death has been reported since 2008, treat that as an upper bound. The machines that tip are almost always responding to applied force from the person who ends up in the fall zone.

The Comparison That Actually Matters

The vending machine statistic gets deployed to make sharks seem safer. The more useful takeaway runs in the other direction: it illustrates how meaningless population-level death counts are when the behavior driving the deaths is not uniformly distributed.

Sharks kill disproportionately among people who swim frequently in coastal waters. Lightning kills disproportionately among outdoor workers in storm-prone regions. Vending machines kill disproportionately among people who apply mechanical force to a top-heavy appliance. Each risk concentrates in a specific behavioral subset, not the national population. The “vending machines vs. sharks” comparison obscures this by treating both numbers as if they reflected similar exposure patterns. The shark statistic already undercounts risk for frequent ocean swimmers, and the vending machine statistic is almost entirely concentrated in a single preventable behavior.

The correct response to learning that vending machines kill two people a year is not “everything is fine.” It is: “Who are those two people, and what were they doing?” The answer, that they were shaking a machine that fell on them, contains the entire risk management lesson. Don’t shake vending machines. Your risk drops to zero.

Sources

• US Consumer Product Safety Commission (CPSC), 1996 news release, “CPSC, Soda Vending Machine Industry Labeling Campaign Warns of Deaths and Injuries”: at least 37 deaths and 113 injuries from 1978 to 1995 from consumers rocking or tilting machines; about 1.7 million soda machines in the marketplace at the time. cpsc.gov
• Slate, “Sharks vs. vending machines: Which kills more people yearly?” (2022): no reported US vending machine death since 2008; CPSC NEISS estimate of about 36,600 emergency department visits associated with vending machines from 2008 to 2021, most minor.
• International Shark Attack File, Florida Museum of Natural History: US unprovoked shark fatalities average about one per year.
• US vending machine count: National Automatic Merchandising Association (NAMA), older estimate of nearly five million machines; 2022–2023 NAMA Industry Census estimates under three million in use.
• Rocking incident count: site estimate of about five million rocking incidents per year (roughly one per machine per year). No agency publishes this figure.
• Machine weight specifications: Standard vending machine weight ranges (400–900 lbs) from industry installation guidelines; fully stocked snack and beverage machines are at the higher end of this range.
• Tip-over mechanism: CPSC incident reports consistently identify rocking/shaking behavior as the precipitating event in fatality cases.