A resident stands at a council meeting with a photograph of her street and asks for the limit to come down from 40 to 30. The engineer is sympathetic and explains that a study was done: a tube across the asphalt, a few hundred vehicles recorded, and a number saying 40 was about right. She leaves believing the town chose 40. The town did not. The drivers did, and the engineer wrote it on a sign.
Most speed limits in the United States reflect no judgment about how fast a street should be travelled. They are set by measuring how fast it already is, taking the speed that 85 percent of free-flowing traffic stays at or below, and posting something close to it.
A rule of thumb from a 1964 study
The reasoning is not stupid. It comes from crash research published by David Solomon in 1964, which found that drivers travelling much faster or slower than surrounding traffic crashed disproportionately often. Fifteen percent of drivers exceed the 85th percentile by definition, so posting a limit where that group separates from everyone else aims enforcement at the genuinely out of step. A limit most people ignore also teaches everyone that signs are decorative.
The trouble came when a rule derived from rural highways in the early 1960s got applied to every street with a curb on it. The National Transportation Safety Board examined this in a safety study adopted in July 2017 and concluded there is no strong evidence that the 85th percentile matches the speed with the lowest crash rate across all road types.
The number that moves when you chase it
The measurement is also not fixed. Raise a limit to match observed speeds and observed speeds rise, producing a higher 85th percentile at the next survey, which justifies raising it again. The safety board named the cycle plainly: speed escalation, with reduced safety as the output.
Texas State Highway 130 is the clean illustration. In 2011 a 45 mile segment went from 70 to 75 mph, a year later to 80, and the toll portion now carries 85 mph, the highest posted limit in the country. Every step rested on an engineering study leaning on the 85th percentile and was defensible alone. The trend was national: 35 states had maximum limits of 70 mph or above in 2012, and 41 did by 2016.
None of this is a rounding error in public health. Between 2005 and 2014, crashes where an officer noted speed as a factor produced 112,580 deaths, 31 percent of all traffic fatalities. Alcohol-impaired driving accounted for 112,948 over the same decade. Only one of those two figures gets a national campaign.
What the 2023 manual changed and what it kept
The safety board recommended striking the 85th percentile guidance from the federal manual outright. What arrived in December 2023, in the 11th edition of the Manual on Uniform Traffic Control Devices, was narrower and more interesting than a refusal. Section 2B.21 now lists six things a speed study must weigh: roadside development and driveway density, transit stops and parking practices, pedestrian and bicycle activity, lane widths and sight distance, at least twelve months of reported crashes, and a review of past studies for trends. Observed speeds are one entry rather than the answer.
| Road | Standing of the 85th percentile since December 2023 |
|---|---|
| Urban and suburban arterials | Must not set the limit on its own, without all six factors considered |
| Rural arterials serving as main streets | Same restriction as urban arterials |
| Freeways, expressways, rural highways | Limit should still land within 5 mph of it, once other factors are found non-mitigating |
One further clause inverts the old logic. Where measured speeds run well above the posted limit and the surroundings do not support a higher one, the study is now directed to consider changing the road: geometry, enforcement, measures that slow traffic. The old instinct moved the sign toward the drivers. The new one moves the drivers toward the sign.
Why five miles per hour is not a small difference
Arguing over 5 mph sounds like fussing until you look at what a body absorbs. The AAA Foundation for Traffic Safety published research by Brian Tefft in 2011 on impact speed and pedestrian outcomes. The average risk of death reaches 10 percent at 23 mph, 25 percent at 32 mph, 50 percent at 42 mph and 75 percent at 50 mph. Between 32 and 42, ten miles per hour doubles the chance a struck person dies. On the street where the survey said 40 and somebody asked for 30, that gap is the whole argument, and it appears nowhere in a spot speed study, which measures vehicles rather than consequences.
Boston moved one number and the fast tail shrank
The usual objection is that a sign cannot change behaviour. Boston tested it. On 9 January 2017 the city dropped its default limit from 30 to 25 mph and publicised the change, and the Insurance Institute for Highway Safety measured speeds before and after against control sites in Providence, where nothing changed. The odds of a vehicle exceeding 25 mph fell 2.9 percent, exceeding 30 mph fell 8.5 percent, and exceeding 35 mph fell 29.3 percent.
That shape is the finding. The ordinary driver barely slowed. What shrank was the tail, and the tail is where the deaths are, since being hit at 30 rather than 38 is roughly a hospital stay rather than a funeral. Thinning the fastest slice while leaving most trips unchanged is a good trade, and it is invisible to anyone reporting average speed.
Reading your own street's speed study
Speed studies are public documents held by a city traffic department, a county engineer or a state transport agency. If a limit near you feels wrong, ask for the one covering your segment and read it with these four questions in hand.
- Check the date. A study predating the school, the apartments or the bike lane describes a street that no longer exists, and changed context is grounds for a fresh one under the current manual.
- Look for the six factors. A study reporting only a speed distribution is running the old method. Ask in writing where crossings, driveways and transit stops were recorded.
- Find the crash history. Twelve months minimum is the guidance. If crashes appear nowhere in the document, the sign was set without reference to anyone getting hurt.
- Ask which road type they claim. The 5 mph rule survives on freeways and rural highways. If your street has houses, shops and people crossing it, that clause does not apply.
Take the answers to the road authority rather than the council, since limits are usually an engineering decision with a statutory basis. A request naming the manual section, the study date and what has changed since is harder to file away than a complaint about fast drivers.
The sign is a decision, not a measurement
The 85th percentile turned a value question into an arithmetic one, which is how it lasted so long. Ask how fast cars should pass a primary school and you get an argument. Ask what speed 85 percent of them stay under and you get a defensible number nobody has to own.
The 2023 manual took some of that shelter away on the streets where people walk. Six factors have to be weighed now, and weighing is a judgment somebody signs. A town still posting numbers from a decade-old tube count runs a method the federal manual has narrowed and the safety board asked to retire, and the street outside the school is where that gap gets paid.