How Rounding Numbers Works — A Complete Guide

A shop sign says a television costs “about £500”. A weather report says the temperature will be “about 21 degrees”. A recipe says a sauce serves “about four people”. None of these numbers is exact, and none of them needs to be. Rounding is the skill of replacing a precise number with a simpler one that is close enough for the job at hand, and it is used far more often in everyday life than exact arithmetic. The rules are short, but they are easy to misapply, especially when decimals are involved.

This guide explains what rounding means, how to round whole numbers to the nearest 10, 100 and 1,000, how to round decimals to a given number of decimal places, and what significant figures are. It works through rounding a crowd size, a decimal measurement and a money total, and it shows where people most often go wrong, so you can round with confidence and know when an exact answer is needed instead.

What rounding means

To round a number is to replace it with a nearby number that has fewer digits or ends in zeros, because the simpler number is easier to use. Rounding 47 to the nearest 10 gives 50, because 50 is the nearest multiple of 10, and 47 is closer to 50 than to 40. The rounded answer is not the same as the original number, and that is the point: it trades a small amount of accuracy for a large amount of simplicity. A rounded answer should always be understood as approximate, and it is often introduced with the word “about” for exactly that reason.

Rounding is useful whenever the exact figure does not matter, or is not even known. Crowd sizes, distances on a road sign, and estimates before a calculation are all naturally rounded, because the exact count would add nothing. It is not useful when the exact figure is the whole point: a bank balance, a medicine dose, or the number of tickets sold must be exact, and rounding them would change the meaning. The first question in any rounding problem is therefore not how to round, but whether rounding is appropriate at all.

Rounding whole numbers

To round a whole number, decide which place you are rounding to — the nearest 10, the nearest 100, or the nearest 1,000 — and then look at the digit just to the right of that place. If that digit is 5 or more, round the chosen place up by one. If it is 4 or less, leave the chosen place unchanged. Every digit to the right of the chosen place then becomes zero. This is the half-up rule, and it is the convention used in school mathematics and in most everyday work: exactly halfway goes up.

Take 47 rounded to the nearest 10. The tens digit is 4, and the digit to its right is 7, which is 5 or more, so the tens digit rounds up to 5 and the units digit becomes zero, giving 50. Take 43 instead: the digit to the right is 3, which is less than 5, so the tens digit stays at 4 and the answer is 40. The halfway case is 45, which rounds up to 50, because 45 is exactly halfway between 40 and 50 and the rule sends halfway cases up. A number line makes this visible: 43 sits nearer 40, 47 sits nearer 50, and 45 sits exactly in the middle, tipped upward by the rule.

The same method works for larger places. To round 847 to the nearest 100, look at the tens digit, which is 4, so the hundreds digit stays at 8 and the answer is 800. But 850 rounds to 900, because the tens digit is 5 and the rule sends it up. To round 2,847 to the nearest 1,000, look at the hundreds digit, which is 8, so the thousands digit rounds up and the answer is 3,000. And 2,449 rounds to 2,000, because the hundreds digit is 4. In each case only one digit decides the answer, and every digit to its right is set to zero, not simply crossed out.

Rounding decimals

Rounding a decimal works the same way, except that the answer is given to a number of decimal places instead of to a multiple of 10. To round to one decimal place, look at the second decimal digit: if it is 5 or more, round the first decimal digit up; if it is 4 or less, leave it unchanged, and drop the rest. So 3.47 rounds to 3.5 to one decimal place, because the second decimal digit is 7, while 3.44 rounds to 3.4, because the second decimal digit is 4. To round to two decimal places, look at the third decimal digit in the same way: 2.675 rounds to 2.68, because the third decimal digit is 5, and 0.125 rounds to 0.13 for the same reason.

Money is the commonest place where decimal rounding matters. Prices are usually given to two decimal places, and a total that works out to more decimals must be rounded to the nearest penny. An amount of £7.895 rounds to £7.90, because the third decimal digit is 5. But rounding should happen at the end of a calculation, not in the middle: adding two amounts that have each been rounded first can give a different total from rounding the exact total once. For example, £4.67 plus £3.48 is exactly £8.15, but rounding each amount to the nearest 10 pence first gives £4.70 plus £3.50, which is £8.20 — five pence too much. Keep the full figures until the last step, and round only the final answer.

Small decimals need the same care with zeros. The number 0.0049 rounded to two decimal places is 0.00, because the third decimal digit is 4, while 0.005 rounds to 0.01, because the third decimal digit is 5. The zeros are part of the answer, not decoration: writing 0 instead of 0.00 would lose the information that the answer was given to two decimal places. Similarly, 1.05 rounds to 1.1 to one decimal place, and the answer is written as 1.1, not 1.10, because only one decimal place was asked for.

Significant figures

Significant figures are another way of saying how much detail a rounded answer should carry, and they are used mainly in science and measurement. The first significant figure is the first non-zero digit, counting from the left. To round to two significant figures, keep the first two significant digits and round using the digit that follows, exactly as before. So 0.00456 rounds to 0.0046 to two significant figures, because the first two significant digits are 4 and 5, and the next digit is 6, which rounds the 5 up. The leading zeros are not significant; they only show where the decimal point falls.

For large numbers the same idea applies, with zeros filling the places that were dropped. The number 34,567 rounded to two significant figures is 35,000, because the first two digits are 3 and 4, the next digit is 5, which rounds the 4 up to 5, and the remaining digits become zeros. And 123,456 rounded to the nearest 1,000 is 123,000, while rounded to the nearest 10,000 it is 120,000 — the same number, rounded to different levels of detail, for different purposes. Significant figures and decimal places answer different questions: decimal places fix how many digits come after the point, while significant figures fix how many meaningful digits the answer carries in total.

Worked example 1: rounding a crowd and a distance

A school fair raises £2,847 and 1,450 people attend. Round each figure to the nearest 100 and to the nearest 1,000 where sensible. For the money, the nearest 100 looks at the tens digit: 2,847 has a tens digit of 4, so it rounds to £2,800. To the nearest 1,000, the hundreds digit is 8, so it rounds to £3,000. For the attendance, the nearest 100 looks at the tens digit, which is 5, so 1,450 rounds to 1,500. To the nearest 1,000, the hundreds digit is 4, so it rounds to 1,000. The rounded figures are easier to report — “about £2,800 raised and about 1,500 people” — but the exact figures are the ones that belong in the accounts.

This example also shows why the level of rounding must match the purpose. Reporting attendance to the nearest 1,000 would hide the difference between 1,450 people and 1,050 people, which matters to the organisers, while reporting it to the nearest person would be false precision, because the count itself was approximate. The nearest 100 is the honest middle ground here, and choosing that level is a judgement call, not a rule of arithmetic.

Worked example 2: rounding decimals and estimating

A piece of wood is measured as 2.675 metres long and another as 1.325 metres. What is their total length, rounded to two decimal places? Adding the exact lengths gives 4.000 metres, which rounds to 4.00 metres. If each length is rounded first, 2.675 becomes 2.68 and 1.325 becomes 1.33, and their sum is 4.01 metres — one centimetre too much. The error is small, but it is exactly the kind of error that matters when pieces must fit together, and it is why rounding is left to the end.

Rounding is also the fastest way to estimate before calculating. To estimate 48 times 6, round 48 to 50 and multiply: 50 times 6 is 300, while the exact answer is 288. The estimate is 12 too high, which is close enough to confirm that an exact answer of 288 is sensible and that an answer of 2,880 would be a slip. Estimates of this kind are worth doing even when a calculator is available, because a calculator will faithfully repeat a wrong key press, while an estimate will expose it.

Worked example 3: the step-by-step trap

Round 3.45 to the nearest whole number. The correct method rounds once, from the original number: look at the first decimal digit, which is 4, which is less than 5, so the answer is 3. The trap is to round in steps: first to one decimal place, giving 3.5, and then to a whole number, giving 4. That answer is wrong, and the mistake is rounding twice. Each rounding step introduces its own error, and the errors compound. The rule is simple: always round from the original, unrounded number, directly to the place you need, in a single step.

The same trap appears with longer decimals. Rounding 7.895 to one decimal place directly gives 7.9, because the second decimal digit is 9. Rounding it first to two decimal places gives 7.90, and then to one decimal place gives 7.9 again, so no harm is done in that case — but that is luck, not method. With 3.45 the two routes disagree, and only the direct route is right. When a problem gives a long decimal and asks for a rounded answer, cover up the digits beyond the deciding one and decide from the original figure alone.

Common mistakes to avoid

Rounding 5 down is the most common error: 45 to the nearest 10 is 50, not 40, and 850 to the nearest 100 is 900, not 800, because the half-up rule sends exactly halfway cases up. The second is rounding in steps, as in the 3.45 example above, which can change the answer by a whole unit. The third is confusing decimal places with significant figures: 0.00456 to two decimal places is 0.00, but to two significant figures it is 0.0046, and the two answers describe quite different levels of detail.

Two more traps are worth watching for. One is rounding too early in a calculation, which lets small errors grow, as the £4.67 and £3.48 example showed. The other is rounding an answer and then presenting it as exact: a rounded figure should carry the word “about” or the symbol ≈, and a final answer in a school problem should state the level it was rounded to, such as “to two decimal places”. A rounded answer without that label invites the reader to treat an approximation as a fact, which is how small errors become confident mistakes.

Frequently asked questions

Why does 5 round up instead of down?

It is a convention, not a law of nature: exactly halfway cases have to go somewhere, and the half-up rule is the one used in school mathematics and in most everyday and financial work. Some specialist systems, such as banker's rounding, send halfway cases to the nearest even number to avoid a systematic upward drift over thousands of calculations, but that is not the convention expected in school problems or on a shop receipt. Unless a problem says otherwise, 5 rounds up.

Does rounding change the answer to a calculation?

Yes, by design. A rounded answer is an approximation, and it will differ from the exact answer by less than half the rounding unit: rounding to the nearest 10 can be wrong by at most 5, and rounding to two decimal places by at most 0.005. That is acceptable when the rounded figure is used for estimating, reporting or comparing, and unacceptable when the exact figure is needed. If a problem asks for an exact answer, do not round at all.

Should I round when I am working with money?

Round only the final total, to the nearest penny, and keep the exact figures until then. Rounding each item first, or each step of a calculation, can move the total by several pence, as the £8.15 and £8.20 example showed. In a shop the till works in exact pence and only the change is rounded, where cash rounding applies, so the same principle holds in everyday life: exact until the end, rounded only for the final figure that is actually paid or reported.

Can I use a calculator to round numbers?

A calculator will give the exact result, and the rounding is still your job: decide the place, look at the deciding digit, and apply the half-up rule. The Scientific Calculator on this site is useful for the multiplication and division that often come before rounding, and the Average Calculator is worth checking when an average needs rounding to a sensible number of decimal places, since averages often come out with far more digits than the data justifies. Whichever tool is used, the deciding digit should be read from the full, unrounded result.

If you want to check a result, try the matching tools on this site: the Scientific Calculator handles the arithmetic before rounding, and the Average Calculator is the usual place where a rounded average is needed.