The rounding calculator above handles the three things people actually mean when they say "round this number": to a set number of decimal places, to a set number of significant figures, or to the nearest 10, 100, 1000 or any other multiple. It also does the part most rounding tools skip entirely — it shows you what happens at exactly one half, where the common rounding rules genuinely disagree with each other.
That disagreement is not a technicality. Round 2.5 to a whole number and you get 3 using the rule taught in school, or 2 using the rule built into most spreadsheets, statistical software and financial systems. Both are correct under their own convention. Arb Digital publishes this tool with all four results side by side because a rounding difference that appears trivial on one number becomes a systematic bias when applied across a million rows.
What This Rounding Calculator Does
Enter a number and choose how you want it rounded. In decimal places mode, the number is rounded to the chosen number of digits after the point. In significant figures mode, the count starts from the first non-zero digit anywhere in the number, so 0.004567 to two significant figures gives 0.0046. In nearest-multiple mode, the number is snapped to the closest multiple of whatever you specify — 10, 100, 1000, or something irregular like 25 or 0.05.
The headline result uses whichever tie-breaking method you select. The grid always shows four results at once: round half up, banker's rounding, always up, and always down. When these four agree, the number is not near a tie and rounding is uncontroversial. When they diverge, you can see immediately by how much and decide which convention your situation requires.
The working panel spells out the decision: which digit was examined, what the discarded remainder was, and why the number moved the way it did. This is the part that turns a rounded answer into something you can explain to someone who disagrees with it.
How to Use It
- Enter your number. Any magnitude and any sign works, including very small decimals.
- Choose the rounding target. Decimal places for money and measurements, significant figures for scientific work, nearest multiple for pricing and estimating.
- Set the precision. The places field controls both decimal places and significant figures; the nearest field controls the multiple.
- Pick a tie-breaking method if you have a house standard. Otherwise leave it on round half up and compare against the grid.
- Read the working line to see exactly which digit made the decision.
The Formula and How It's Calculated
Rounding to decimal places works by scaling. To round to d decimal places, multiply by 10 to the power d, apply the rounding rule to the resulting whole number, then divide by the same factor. Rounding 3.14159 to two places means multiplying by 100 to get 314.159, rounding to 314, and dividing back to 3.14.
Significant figures use the same mechanism after locating the first significant digit. Work out the magnitude of the number, subtract it from the figure count, and that gives the number of decimal places to round to. For 0.004567 to two significant figures, the first significant digit sits in the third decimal place, so the number is rounded to four decimal places: 0.0046. Nearest-multiple rounding divides by the multiple, rounds to a whole number, and multiplies back — so 1265 to the nearest 10 becomes 126.5, which rounds to 127 or 126 depending on your tie rule, giving 1270 or 1260.
The tie case is where the rules split. With the default value of 2.5 rounded to zero decimal places, the discarded part is exactly one half. Round half up gives 3, because ties go away from zero. Banker's rounding gives 2, because ties go to the nearest even number and 2 is even. Always up gives 3 and always down gives 2. Every method agrees on 2.4 and on 2.6; they only part company at exactly 2.5. The IEEE 754 floating-point standard specifies round half to even as the default behaviour for computer arithmetic, and the NIST Guide for the Use of the International System of Units discusses rounding practice in the context of reporting measured values.
Banker's Rounding and Why It Exists
Round half up has a hidden defect: it is biased upward. Across many numbers ending in exactly 5, every single one moves up, so a long column of rounded values totals slightly more than the unrounded data. On a handful of numbers this is invisible. On a payroll run, an interest calculation across a million accounts, or a scientific data set of any size, the accumulated drift is real and one-directional.
Round half to even removes the bias by sending ties up half the time and down half the time, based on whether the preceding digit is odd or even. Rounding 0.5, 1.5, 2.5 and 3.5 to whole numbers gives 0, 2, 2 and 4 — total 8, exactly matching the unrounded total of 8. Under round half up the same four values give 1, 2, 3 and 4, totalling 10. That two-unit overstatement is the bias, visible in miniature.
This is why banker's rounding is the default in IEEE 754 floating-point arithmetic, in Python's round function, in many statistical packages, and in a great deal of accounting software. It is also why a spreadsheet total can differ from a hand-calculated one by a few pennies, and why exporting data between two systems using different conventions produces reconciliation differences that are maddening to trace. If you cross-check results between tools, our significant figures calculator and this page will both tell you which rule was applied.
Decimal Places Versus Significant Figures
These two are not interchangeable, and the difference matters most for very small and very large numbers. Decimal places count digits after the point regardless of what those digits are. Significant figures count meaningful digits starting from the first non-zero one, wherever it appears.
Round 0.004567 to two decimal places and you get 0.00 — the entire number vanishes. Round it to two significant figures and you get 0.0046, preserving what you actually measured. Conversely, rounding 12,345 to two decimal places changes nothing, while rounding it to two significant figures gives 12,000. Scientific and engineering work uses significant figures precisely because it tracks precision rather than position, which is what matters when your quantities span many orders of magnitude.
The practical rule: use decimal places when the unit is fixed and meaningful, as with currency, where two places is the unit itself. Use significant figures when the magnitude varies and precision is what you are communicating. Presenting more significant figures than your measurement supports implies an accuracy you do not have, which is the false-precision problem our percent error calculator discusses in a measurement context.
Rounding Twice: The Error That Compounds
Never round the same number more than once. Take 2.44 and round it to one decimal place to get 2.4, then round that to a whole number and you get 2. Round 2.44 directly to a whole number and you get 2 as well — fine here. But take 2.45: rounding to one place gives 2.5, then rounding again gives 3, whereas rounding 2.45 straight to a whole number gives 2. The intermediate step manufactured a tie that did not exist in the original data.
This is called double rounding, and it is a genuine source of discrepancies in reporting pipelines where each stage rounds its output before passing it on. The fix is to carry full precision through every intermediate calculation and round exactly once, at the point of presentation. The same principle applies to any multi-step calculation: our long division calculator and percentage calculator both work at full precision and round only the displayed answer for this reason.
Rounding Negative Numbers and Money
Negative numbers expose the difference between the methods that ties do not. Ceiling means "toward positive infinity", so the ceiling of −2.5 is −2, which is arithmetically larger. Floor means "toward negative infinity", so the floor of −2.5 is −3. Truncation is different from both: it simply discards the fractional part, so truncating −2.5 gives −2, matching ceiling for negatives and floor for positives. If your requirement is "always round away from zero", none of the three does that on its own — you need round half up with sign awareness, which is what this calculator applies.
Money adds its own conventions. Prices are frequently rounded to the nearest 5 or 10 cents where small coins have been withdrawn from circulation, which is nearest-multiple rounding with a multiple of 0.05 or 0.10. Tax calculations often specify a rounding rule in legislation, and it is not always round half up. Pricing psychology pushes in yet another direction, with retailers rounding down to .99 regardless of what the arithmetic suggests — our product pricing calculator works through the margin implications of that. Always check whether a rounding rule is mandated before choosing one.
Arb Digital maintains a free library of calculators, converters and checkers. Browse the collection, or tell us what you keep looking for and cannot find.
Browse All Free Tools Suggest a ToolCommon Mistakes to Avoid
- Rounding more than once. Each intermediate rounding can create a tie that did not exist, pushing the final answer a full unit away from the truth.
- Assuming everyone uses round half up. Spreadsheets, programming languages and accounting systems frequently default to banker's rounding, and the totals will differ.
- Confusing decimal places with significant figures. Rounding 0.004567 to two decimal places destroys the number; to two significant figures it preserves it as 0.0046.
- Using floor or truncation for negatives when you mean away from zero. They behave differently on the negative side of zero, and the results differ by a whole unit.
- Reporting more digits than your data supports. Rounding cannot add precision, and quoting extra figures implies an accuracy the measurement never had.
Related Free Tools From Arb Digital
For precision in scientific reporting, use the significant figures calculator and the scientific notation converter. To keep measurement accuracy honest, use the percent error calculator. The long division calculator and fraction calculator produce the recurring decimals you will most often need to round, and the percentage calculator handles shares and proportions. Browse the full free online tools hub for more.
Frequently Asked Questions
Both answers are correct under different conventions. Round half up, the rule taught in most schools, gives 3 because ties move away from zero. Banker's rounding gives 2 because ties move to the nearest even number, and 2 is even.
Banker's rounding, also called round half to even, sends a value ending in exactly one half to whichever neighbouring option is even. It exists to remove the upward bias of always rounding ties up, which matters when rounding many values and summing them.
Decimal places count digits after the point regardless of value. Significant figures count meaningful digits from the first non-zero digit onward. Rounding 0.004567 to two decimal places gives 0.00, while two significant figures gives 0.0046.
Divide by the multiple, round the result to a whole number, then multiply back. Rounding 1265 to the nearest 10 means dividing by 10 to get 126.5, rounding that to 127 or 126 depending on the tie rule, and multiplying back to 1270 or 1260.
Because they may use different tie-breaking conventions. Many spreadsheets, programming languages and accounting systems default to banker's rounding, while handheld calculators and school teaching usually use round half up.
Yes, in almost every case. Rounding 2.45 to one decimal place gives 2.5, and rounding again gives 3, whereas rounding 2.45 straight to a whole number gives 2. Carry full precision through the calculation and round only once at the end.
Ceiling moves toward positive infinity, so the ceiling of minus 2.5 is minus 2. Floor moves toward negative infinity, giving minus 3. Truncation simply discards the fraction, giving minus 2. Round half up moves away from zero, also giving minus 3.
Rounding rules for tax, payroll and financial reporting are often specified by law or accounting standard. This page explains how the arithmetic works and is not financial or tax advice.
