Rounding Calculator - Decimal Places, Significant Figures, Place Value

A rounding calculator takes an exact (or at least fully written) decimal and shortens it to a requested precision: a count of decimal places, a count of significant figures, or the nearest place value such as 10, 100, 1000, 0.1, or 0.01. The point is not to invent accuracy. The point is to stop writing digits that your measurement, your assignment, or your report does not claim.

Schools, labs, and finance disagree about what to do when the leftover is exactly five. IEEE-754 default rounding in most CPUs is round half to even (banker’s rounding). Many grade-school texts teach “5 and above, round up” without stating a sign rule. This free online rounding tool documents round half away from zero: 2.5 becomes 3 and −2.5 becomes −3. You can switch to half-up (toward +infinity) if that is what your course uses. It does not apply banker’s rounding.

The default example is the first digits of π: 3.14159 to 3 decimal places is 3.142. The fourth decimal digit is 5 and the digits after it are not all zero (the last digit is 9), so the leftover is greater than half a thousandth and the third decimal place rises from 1 to 2. The result panel always shows the original, the rounded value, and how many digit positions changed so you can check a homework step rather than only copying a final answer.

Everything runs in the browser. Your number is not uploaded. Pair this page with the scientific notation calculator when you need to show significant figures as a coefficient times a power of ten, with the percentage calculator when a rounded percent must stay consistent with a raw ratio, or with the scientific calculator when the value you are rounding is the output of a longer expression.

Using this decimal-place and significant-figure rounder is straightforward:

1. Enter the original number. Accepts 3.14159, −2.50, 1,234.5, and short scientific form such as 1.2e3. Total digits are capped at 40 so the display stays readable.
2. Choose how to round: decimal places from 0 through 12, significant figures from 1 through 15, or a place value (nearest 10, 100, 1000, 0.1, or 0.01).
3. Choose the half rule. Keep half away from zero unless your teacher specified half-up. The two rules agree on every positive example and disagree on negative exact-halves.
4. Read three outputs: original, rounded, and digits changed. Copy the rounded value, or clear the field to start over. The last settings are saved locally for up to 30 days.

Note: rounding uses the decimal digits you typed, not a binary float approximation of those digits. Exponents in E-notation are limited to ±12.

Decimal places count digits after the decimal point. Zero decimal places means the nearest integer. Twelve decimal places is enough for most homework and for money written to tenths of a nanounit; it is not arbitrary-precision arithmetic.

Worked cut at 3 decimal places

• Write 3.14159 with the cut after the 1: 3.141 | 59
• The first discarded digit is 5, and 9 remains, so the leftover is more than half
• Increase the last kept digit: 1 becomes 2
• 3.142 - this is the default on this page

If you stop at 2 decimal places instead, the cut is 3.14 | 159. The first discarded digit is 1, which is less than 5, so the result stays 3.14. Rounding is not “round step by step through every extra digit”: you do not round 3.14159 → 3.1416 → 3.142 → 3.14. You look once, at the first discarded position, and you keep a sticky record of whether anything after a 5 is nonzero.

Negative values use the same leftover test. Under half away from zero, −3.14159 to 3 d.p. is −3.142: the magnitude grows. Under half-up, a negative exact half would shrink in magnitude instead (toward +infinity).

Significant figures start at the first non-zero digit and count every measured digit after that, wherever the decimal point sits. Decimal places ignore magnitude and only count the fractional tail. The two requests are not interchangeable, which is why this calculator has separate modes.

Leading zeros after a decimal point are not significant: in 0.004162 they only locate the 4. Trailing zeros in 4.20 are significant if they were measured. Integer trailing zeros in 42300 are ambiguous on paper; scientific notation (4.23 × 10⁴ versus 4.2300 × 10⁴) removes the ambiguity. Use the scientific notation calculator when your lab report needs that form.

Chemistry and physics courses often ask for 2 or 3 significant figures after a multiplication. Money and sports times usually ask for decimal places (cents, hundredths of a second). If the problem says “nearest hundredth,” that is 2 decimal places, not 2 significant figures.

Place-value rounding means “nearest multiple of this unit.” Nearest 10 is the same idea as −1 decimal places; nearest 0.01 is the same idea as 2 decimal places. This mode exists so you can pick the unit by name instead of converting it into a decimal-place count.

• 37 to the nearest 10 is 40
• 1234 to the nearest 10 is 1230; nearest 100 is 1200; nearest 1000 is 1000
• 3.14159 to the nearest 0.1 is 3.1; nearest 0.01 is 3.14
• Ties use the same half rule as decimal-place mode

Census tables, map scales, and “round to the nearest dollar” are place-value problems. “Nearest dollar” is nearest 1, which you can do with 0 decimal places. This calculator’s place-value list stops at thousands and hundredths because those are the units that show up constantly in school exercises; it does not offer nearest million or nearest millionth.

An exact leftover of one-half is the only case where reasonable people disagree. This page’s documented default is round half away from zero: the magnitude of the number always increases on a tie. Half-up, in the Java RoundingMode.HALF_UP sense, rounds toward +infinity: positive ties grow, negative ties shrink.

If discarded digits after a 5 include any non-zero, the leftover is greater than half and both rules round away from zero. That is why 3.14159 to 3 d.p. is 3.142 under either half option. Banker’s rounding (2.5 → 2, 3.5 → 4) is not offered; say so in a lab report if your software used it.

When a problem asks for a fixed number of decimal places, the trailing zeros are part of the answer: 3.1 to 3 d.p. is 3.100, not 3.1. This calculator keeps those zeros in decimal-place mode so a printed result matches a “three decimal places” rubric. In significant-figure mode, extra zeros in 1200 from rounding 1234 to 2 s.f. are placeholders for the tens and ones; they do not mean you measured four digits.

Rounding cannot recover information you already threw away. If a spreadsheet stored π as 3.14, rounding that 3.14 to 5 decimal places still yields 3.14000 - five digits of display, two digits of information. Feed the calculator the most precise original you actually have.

A rounding error on a test is almost always one of four habits: rounding twice, mixing decimal places with significant figures, forgetting the sign rule on a 5, or treating a display width as new information. Walk through each with the same original 3.14159 so the modes on this page stay distinct.

Do not round twice

Students sometimes round 3.14159 to 3.142 (3 d.p.) and then round that 3.142 to 3.14 (2 d.p.). The second step looks at a 2, which would keep 3.14 - accidentally correct here - but the method is wrong. The legal 2-decimal-place cut looks at the original third digit 1 and also yields 3.14. Change the original to 1.2496. Rounded once to 3 d.p. it becomes 1.250; rounding that again to 2 d.p. becomes 1.25. Rounded once from the original to 2 d.p. it becomes 1.25 as well in this case, but 1.24949 to 2 d.p. is 1.25 if you wrongly go through 1.249 → 1.25 after a 3 d.p. stop, versus 1.25 directly… wait: 1.24949 to 2 d.p. looks at 9, so 1.25 either way. The famous trap is 1.249 to 2 d.p.: the third digit is 9, so the answer is 1.25. If someone first “rounds to 3 d.p.” they still have 1.249, then a second round still sees 9. The cleaner trap is 2.446 rounded stepwise 2.45 then 2.5 when a single 1-decimal-place cut should have looked at 4 and stayed 2.4. Always cut once from the original.

Spreadsheets are not this page

Excel ROUND uses half-up toward +infinity for binary-converted decimals, which is already a different object than “the decimal digits I typed.” ROUNDUP always moves away from zero. ROUNDDOWN always toward zero (truncation). MROUND rounds to a multiple. Python 3 round(2.5) is banker’s rounding and returns 2. JavaScript Math.round(2.5) is half-up for positives (3) and is not a decimal-place routine until you scale. This calculator is a decimal-digit rounder with a labeled half rule. If your lab partner’s Python notebook disagrees on an exact .5, check whether they used banker’s rounding before you “correct” your homework.

More cuts of 3.14159

• 0 d.p. (nearest integer): 3
• 1 d.p.: 3.1 · 2 d.p.: 3.14 · 3 d.p.: 3.142 · 4 d.p.: 3.1416
• 2 significant figures: 3.1 · 4 significant figures: 3.142
• Nearest 0.1: 3.1 · nearest 0.01: 3.14 · nearest 10: 0

Nearest 10 of 3.14159 is 0 because 3 is closer to 0 than to 10 (the midpoint would have been 5). That surprises people who think “place value” only applies to large whole numbers. It applies to any real you can write. The digits-changed counter on the result panel will show several positions flipping when you jump from a 6-digit decimal to a multiple of 10; that is expected, not a bug.

Percentages compound the vocabulary. “Round the percentage to 1 decimal place” is a decimal-place request on the already computed percent, not a significant-figure request on the raw count. Compute the percent first (see the percentage calculator), then round that output here. Reversing the order - rounding the count, then dividing - is a different number and is wrong unless a rubric says to round intermediates.