Rounding

Rounding to Decimal Places

Rounding to the nearest hundredth means keeping two decimal places, and the digit in the third place decides whether the second one goes up or stays put. This guide names every place properly, covers money, negatives and carries, and includes a chart of numbers rounded four ways at once.

Rounding to the nearest hundredth, and what the places are called

The hundredths place is the second digit after the decimal point. To round there, look at the third digit: 5 or more and the second digit goes up by one, 4 or less and it stays as it is. Everything past that point is discarded. So 3.14159 becomes 3.14 and 2.71828 becomes 2.72.

Decimal places have names, and the names are just place value continued to the right of the point. Take 4.5678:

• The tenths place is the 5. Rounding to one decimal place is rounding to the nearest tenth, and 4.5678 becomes 4.6.
• The hundredths place is the 6. Two decimal places is the nearest hundredth, and 4.5678 becomes 4.57.
• The thousandths place is the 7. Three decimal places is the nearest thousandth, and 4.5678 becomes 4.568.
• The ten-thousandths place is the 8, and four decimal places leaves the number as 4.5678.

Each place is a tenth the size of the one before it, so each extra decimal place is ten times more precise. Zero decimal places means the nearest whole number, and 4.5678 becomes 5.

Only one digit gets a vote

This is where most errors come from. When rounding to the nearest whole number, 3.4999 becomes 3, not 4. The deciding digit is the 4 in the tenths place, and the three 9s behind it are irrelevant no matter how close to 3.5 they push the number. You never look at two digits, and you never round the tail first and work backwards. The rounding calculator highlights the deciding digit for whichever place you pick, which makes the habit easier to build.

Rounding chart: the same numbers at 0, 1, 2 and 3 decimal places

Every row below is one number rounded four different ways using the standard rule. Reading across a row shows what each level of precision costs; reading down a column shows how the rule behaves on awkward numbers.

Several rows are worth a second look, because they are the ones that make people think the rule is broken.

• 0.08512 rounds to 0.1 at one decimal place but 0.09 at two, which looks contradictory and is not. At one place you keep 0.0 and the deciding digit is the 8, so the tenths digit goes from 0 to 1. At two places you keep 0.08 and the deciding digit is the 5, so the 8 goes to 9. Different question, different deciding digit.
• 7.0649 gives 7.1 at one place and 7.06 at two. The 1 dp answer is decided by the 6, not by the 06 taken together.
• 8.04999 gives 8.0 at one place and 8.05 at two, for the same reason in reverse.
• 9.99951 carries all the way up at every level, turning into 10, 10.0, 10.00 and 10.000.
• −0.0637 rounds to 0 at zero decimal places. Some software displays that as −0, which is the same value written oddly.

The trailing zeros in the table are deliberate. 1.970 and 10.000 are written out in full because a value quoted to three decimal places should show three decimal places, even when the last ones are zeros.

When rounding carries: 1.97 becomes 2.0

Rounding up a 9 has nowhere to go, so it carries into the place to its left, exactly as it does when you add.

The chain can run further. 9.999 to two decimal places is 10.00, because the carry passes through the hundredths, the tenths and the units in turn. If the digits ahead of the carry are all 9s, the whole number rolls over.

Keep the trailing zeros

Writing the answer to 1.97 as “2” is wrong in a context that asked for one decimal place. The zero is not decoration; it states that the value is precise to the tenths place. The same applies at two places:

In a column of figures this matters. A table where some values read 2.0 and others read 2 invites the reader to assume the second kind was measured less carefully, and spreadsheets strip those zeros unless the cell format is fixed to the number of places you want.

How to round to the nearest cent

Rounding to the nearest cent is rounding to the nearest hundredth, because a cent is one hundredth of a dollar. Do it once, on the final figure, after all the arithmetic is finished.

Sales tax rates differ by state and often by city or county, so 8.25% here is only an example rate, not a national figure. Check the rate that applies where the sale happens. Fuel prices are the well-known case of a price quoted in thousandths: the pump advertises $3.499, and the total you pay is rounded to the cent at the end.

Splitting money that will not divide

Some amounts cannot be split evenly, and rounding each share is not enough on its own.

The stray cent has to go somewhere. Give it to one share, or spread the leftovers deliberately, but do not pretend the rounded shares add back to the total. The same drift appears across an invoice: round each line to the cent and the sum will usually differ from the tax on the unrounded total by a cent or two.

Rounding negative numbers

For any number that is not sitting exactly on a halfway point, rounding a negative is the same as rounding its positive twin and putting the minus sign back.

Notice that −4.5678 becomes −4.6, a number further from zero than −4.5. That is correct, and it is also the point where the words start working against you. “Rounding up” on the number line means moving toward positive numbers, so rounding −4.5678 up would give −4.5. What actually happened is that the standard nearest-value rule chose −4.6 because it is nearer.

Exact halves are the only place the conventions genuinely disagree, and it is worth knowing which one you are using:

• Round half up, read literally as toward the larger value, sends −2.345 to −2.34.
• Round half away from zero sends −2.345 to −2.35, mirroring what happens to +2.345.
• Round half to even sends it to −2.34, because 4 is even.

Away from zero is the convention most people expect, since it treats a debit and a credit of the same size identically. Whichever you pick, apply it to the whole column.

The double-rounding trap

Rounding in stages can give a different answer from rounding once. This is the failure that hides inside neat step-by-step working, and it is the reason to keep full precision until the last line.

The first answer is the right one. 2.4449 is below 2.445 and therefore closer to 2.44, but rounding to three places promotes it onto a halfway mark it was never on, and the second round then pushes it up. The longer the chain, the worse it gets: 1.4449 rounds directly to 1, while rounding place by place gives 1.445, then 1.45, then 1.5, then 2.

In money this shows up constantly. A unit price of $2.4449 is $2.44 to the cent. Round it to $2.445 in an intermediate column and the final figure becomes $2.45, and on a thousand units that is a $10 discrepancy created purely by the working.

The rule is short: carry every digit you have through the calculation, and round exactly once, straight from the original value to the place you actually need. When you are averaging, take the full mean from the average calculator and round that, rather than rounding each value first and averaging the rounded ones. If you need to state how far a rounded figure sits from the true one, the percent error calculator turns that gap into a proportion.

Decimal places and significant figures are not the same thing

Decimal places count digits after the point. Significant figures count meaningful digits starting from the first non-zero one, wherever that happens to sit. They answer different questions, and on the same number they can give wildly different results.

Leading zeros are never significant; they only mark where the number sits on the scale. Trailing zeros after a decimal point are significant, which is why 0.004520 has four significant figures rather than three.

Which to use depends on what fixes your precision. Money has a fixed scale, so decimal places are the right unit: every amount is quoted to the cent whether it is $3.00 or $3,000,000.00. Measurements that span orders of magnitude are better handled with significant figures, since three significant figures means the same relative precision at 0.00782 as at 78,200. For that job the significant figures calculator counts and rounds them correctly, including the awkward trailing-zero cases.

At the far end of the scale, decimal places stop being the useful question entirely. When a value runs to dozens of digits, rounding is what introduces the error rather than what controls it, and the big number calculator keeps the arithmetic exact instead.

Common mistakes

• Reading past the deciding digit. Only the first digit after the cutoff matters. 3.4999 rounds to 3 at zero decimal places.
• Rounding in stages. 2.4449 is 2.44 to two places, but 2.45 if you stop at three places on the way. Round once, from the original.
• Dropping the trailing zero. 1.97 to one decimal place is 2.0. Writing 2 claims a different level of precision.
• Rounding each line, then adding. The rounded total will not match the total of the unrounded values. Add first and round the sum.
• Assuming the rounded shares add back to the whole. $100 split three ways gives $33.33 each and leaves a cent over. Somebody has to get it.
• Mixing up the two negative conventions. −2.345 is −2.34 under half up and −2.35 under half away from zero. State the rule before you build the column.
• Confusing decimal places with significant figures. 12345.6 to two decimal places is 12345.60; to two significant figures it is 12000.
• Trusting a displayed value. What a spreadsheet shows and what it stores are often different numbers, which is why totals can look off by a cent.
• Reporting more precision than you measured. Rounding never adds accuracy. Quoting four decimal places on a figure measured to two overstates what you actually know.

Frequently asked questions

Calculators for this

Every one of these shows the working, so you can check the method as well as the answer.

More guides