The rule
First decide which digit is to be the last one. Then look at the digit immediately after it:
- If it is 0, 1, 2, 3 or 4 — leave the last digit as it is.
- If it is 5, 6, 7, 8 or 9 — increase the last digit by one.
Every digit after the last one is dropped. The rule works the same for decimals, tens and millions, because it is the same place-value system on both sides of the decimal point.
3.847 to two decimals: the last digit is to be the 4, the next one is 7, so 3.85.
3.847 to one decimal: the last is to be the 8, the next is 4, so 3.8.
The two do not agree. Rounding in two steps — first to 3.85 and then to 3.9 — gives the wrong answer. The rounding is done in one step, from the original number.
Where the rounding belongs
Last. Rounding throws information away, and information thrown away does not come back at the next step. Work all the way through with the numbers you were given, and round only as the answer is written down.
An intermediate result rounded at every step can drift noticeably from the right one, and the drift can grow with the number of steps. That is why a calculator holds more decimals than it shows.
Tens, hundreds and whole units
The same rule, a different last digit:
- 1847 to the nearest hundred: the last digit is the 8, the next is 4, so 1800.
- 1872 to the nearest hundred: the next digit is 7, so 1900.
- 47.5 to a whole number: the next digit is 5, so 48.
Rounding an amount to whole units is the same operation: 129.49 becomes 129 and 129.50 becomes 130.
Worked examples
Two decimals. 0.12563 → 0.13.
One decimal. 9.96 → 10.0. The last digit became 10, so a one is carried — and the zero is written out, because it says the answer holds to one decimal.
Tens. 2345 → 2350.
Up despite a low digit? No. 5.049 to one decimal is 5.0, not 5.1 — it is the digit immediately after that decides, not the ones beyond it.
Common mistakes
- Rounding in several steps. 3.847 → 3.85 → 3.9 is wrong; the answer is 3.8.
- Looking at more than one digit. Only the digit immediately after decides.
- Dropping a trailing zero. 10.0 to one decimal is not the same as 10.
- Rounding early in a calculation. Round last, work exactly until then.
Common questions
What happens with exactly 5?
School mathematics usually rounds up: 2.5 becomes 3. Some fields use a different rule that rounds to
the nearest even digit so that errors do not accumulate, but that is not the school rule.
How do you round negative numbers?
Round the size and keep the sign: −3.847 to two decimals is −3.85. “Up” then means away from zero.
How is this different from significant figures?
Rounding is about how many decimals are written out.
Significant figures are about how many
digits the number claims to know, which is a different question for very large and very small numbers.