GMAT® Decimal Rounding Rules: A Complete Guide

Rounding is old material, and it makes sense to treat it that way. You rounded to the nearest ten in elementary school, you round tips and grocery prices as an adult, and a question built on that skill can look like free points on a test that rarely gives any away.

But the GMAT® rarely asks for the skill in the form you practiced it. The exam makes its move in two places the elementary version never touched: when the rounding has to happen once and only once, and when a rounded number arrives as the input rather than the answer. Both show up on questions where the arithmetic stays simple and the trap does the work. So it's worth a few minutes to rebuild the rule on purpose, see where it bends, and then look at what the test does with it.

The rule itself

To round a number to a given place, look at the single digit immediately to the right of that place. A digit of 5 or higher rounds the kept digit up. A digit of 4 or lower leaves it alone, and everything past the digit you kept drops away.

$3.47 \rightarrow 3.5 \qquad 3.44 \rightarrow 3.4$

Both decisions used only one digit each. The hundredths digit decided, and the digits after it don't get a vote, because everything they add together is smaller than one unit of the deciding digit's place. That's what makes the one-look rule exact rather than approximate: a 4 followed by a parade of 9s is still a 4, so $3.44999$ to the nearest tenth is $3.4$, even with the trailing 9s pulling the number toward the midpoint.

Round once, at the end

Rounding as you go feels careful. Each step keeps the numbers manageable, and nothing about it looks reckless on a scratch pad.

The mistake to watch for isn't choosing the wrong direction. It's rounding twice. Take $0.4849$, and round it to the nearest hundredth. Done correctly, the answer is $0.48$, because the hundredths digit is 8 and the digit next to it is 4. But watch what happens when the work passes through a middle step: round $0.4849$ to the nearest thousandth first and you get $0.485$, and now rounding that to the hundredth gives $0.49$, because that fresh 5 tells you to round up. The staged answer and the direct answer disagree, and the direct one is right.

Each rounding throws away information, and an intermediate rounding throws away exactly the information the final decision needed. The 9 in the ten-thousandths place was the only thing proving that $0.485$ was really $0.4849$ in disguise, and the first rounding erased it. So carry full precision through your work, and round once, from the original number, at the end. Multi-step problems on this exam are short enough that the cost of keeping an extra digit or two is trivial, while the cost of a staged rounding error is the whole question.

Compute one more place than you need

Long division stops where you stop it. If you quit at $2.8$ and the question wants the nearest tenth, you're stuck, because a quotient that starts $2.8$ could be $2.80$ or $2.89$, and those two round differently. The deciding digit exists, but your division hasn't produced it yet. Our Math Basics series has a full lesson on the process if the division itself is the wobbly part.

So when a rounding decision hangs on a digit your division hasn't produced, keep going one more step. One extra long-division cycle hands you the deciding digit, and the up-or-down question settles on the spot. A lot of the quotients this exam asks for don't terminate neatly, which makes one-more-place the default posture rather than an occasional rescue. If the division does terminate, the extra step is quick and simply confirms the value.

Reading a rounded value backwards

The harder direction runs the other way: the exam hands you a rounded number and asks about the value underneath.

"290 miles, rounded to the nearest 10 miles" doesn't mean Cindy drove 290 miles. It means her true distance rounds to 290, so it lives somewhere from 285 up to (but not including) 295. A distance of 285 rounds up to 290, and a distance of 294.999 still rounds down to 290, so both edges feed the same rounded value. The rounded number is a label on a band, not a point.

In Data Sufficiency, a statement like "x, rounded to the nearest tenth, is 2.3" tells you that x sits somewhere from 2.25 up to (but not including) 2.35. That's a range with real width, and questions are built on what's still open inside it: whether x tops 2.3, whether it tops 2.5, whether it can be an integer. Naming the band is the move, because the band is all the statement ever gave you.

When two rounded values feed one fraction, the band logic stacks, and the fraction's extremes come from cross-pairing the edges rather than dividing max by max. Our "Cindy drove her car 290 miles, rounded to the nearest 10..." worked solution applies that band logic to a fraction built from two rounded values.

Edge cases the exam likes

Rounding up can carry. Round $0.999$ to the nearest hundredth and the answer is $1.00$, because the 9 in the thousandths place bumps the hundredths digit past 9, and the carry climbs straight through the decimal point. Numbers near a roll-over boundary are where the mechanical rule pays for itself, since the correct answer looks nothing like the number you started with.

When the deciding digit is exactly 5, the number rounds up, so $2.35$ to the nearest tenth is $2.4$. That's the standard convention, and the half-way cases on the exam follow it.

Where rounding turns into estimation

Everything above treats rounding as the skill being tested. Run the same skill forward on purpose and it becomes estimation: round $198$ up to $200$ to size a product, and round one value down when you rounded another up so the errors partly cancel. Whether an estimate is safe is decided by the answer choices: when they're spread apart, a rough answer can eliminate several of them before real computation starts, and when they sit close together, the question probably wants the full solution. Our mental math guide covers that decision in depth, along with the written-computation habits that take over when estimation won't.

Try this one

Try this one before reading on.

The hundredths digit is 8, and the digit next to it is 4, so the 8 stays and the answer is $0.48$, choice (B).

The trap is (D), and it catches the staged version of the same work: round to the nearest thousandth first, land on $0.485$, then round that up to $0.49$. Staging hides the ten-thousandths digit, and that digit was the whole case, because $0.4849$ sits below the $0.485$ midpoint. One look at the original number settles it in a single step.

FAQ

What are the rounding rules on the GMAT®?

To round to any place, look at the single digit immediately to the right of it. A digit of 5 or higher rounds the kept digit up, a digit of 4 or lower leaves it alone, and the digits after it don't change the decision. Round once, from the original number, at the end of your work.

Does 5 round up or down on the GMAT®?

Up. When the deciding digit is exactly 5, the number rounds up, so $2.35$ to the nearest tenth is $2.4$. That's the standard convention.

What does "rounded to the nearest tenth" mean?

A rounded value stands in for a range. A measurement reported as 2.3 could be anything from 2.25 up to (but not including) 2.35, because everything in that band rounds to 2.3. The GMAT® uses this to ask questions about the range underneath a rounded value.

Can you round in stages on GMAT® questions?

Rounding in stages can produce a different answer than rounding once, because each stage discards information the next decision needed. The number $0.4849$ rounds to $0.48$ in a single step, but staged rounding lands on $0.49$. Carry full precision through the work and round once at the end.

How do you round a non-terminating decimal?

Keep dividing one decimal place past the place the question asks for, then apply the rule to the digit you find. Stopping the long division early leaves the deciding digit unknown, and one more cycle settles the up-or-down question.

How does rounding show up on the GMAT®?

The rule shows up directly in some questions, runs estimation in others, and powers the reverse questions that hand you a rounded value and ask about the band underneath it. The reverse version is where the difficulty tends to hide.

Want to learn even more?

Related reading:

• GMAT® Number Properties: A Complete Guide
• GMAT® Divisibility Rules: A Complete Guide
• GMAT® Least Common Multiples: A Complete Guide
• GMAT® Data Sufficiency: How to Approach DS Questions
• GMAT® Mental Math: How to Handle Quant Without a Calculator
• "Cindy Drove Her Car 290 Miles, Rounded to the Nearest 10..." GMAT® Worked Solution

The Math Basics series rebuilds decimal arithmetic from the ground up, including "GMAT® Focus Edition Math Basics: Decimal Long Division" (Lesson 10 in the series). And Episode 31 of our Real GMAT® Problems series, "Percents, Exponents, and Rounding", works the miles-per-gallon trap on audio. You can hear both on Spotify, Apple Podcasts, or YouTube.