How the rounding calculator works
Rounding replaces a number with a value at a chosen level of detail. The answer depends on the number, the place or multiple you choose, and the rounding rule. Standard rounding chooses the nearest value and sends exact halfway cases away from zero: 2.5 becomes 3, while −2.5 becomes −3.
- 1Enter the number. Use a positive number, a negative number or zero. Keep all the original digits so the calculator can round directly to your target.
- 2Choose the target. Select a place from Thousands through Thousandths. Use Custom for a nonnegative whole number of decimal places or a positive multiple such as 5, 0.25 or 0.05.
- 3Choose the rounding rule. Leave Standard selected for nearest-value rounding with ties away from zero, or choose one of the other eight rules described below.
The result gives the rounded number, its change from the original and the bounding multiples. The number line and worked steps explain the choice. A comparison table shows the original number rounded to each preset place, so you can compare levels of detail without rounding an already rounded answer.
The rounding formulas
The first equation applies to a nonnegative whole number of decimal places. For a custom multiple, use that multiple as Step instead. Round means rounding to an integer under the selected rule.
Two decimal places use a step of 0.01; zero decimal places use a step of 1. The tens, hundreds and thousands presets use steps of 10, 100 and 1,000. A custom multiple follows the same divide, round and multiply procedure: 17.3 ÷ 5 = 3.46, which rounds to 3 under Standard, then 3 × 5 = 15.
Floor selects the greatest integer at or below a value; ceiling selects the least integer at or above it. If the number is already a multiple of Step, Lower and Upper both equal the number.
A positive change means the answer is numerically larger; a negative change means it is smaller. Under any nearest-value rule, the distance moved cannot exceed half the step. Directed rules such as ceiling and floor can move farther because they do not necessarily choose the nearest value.
Worked example: rounding to the nearest hundredth
Enter 1,234.5678, choose Hundredths and use Standard. The hundredths digit is 6, and the next digit is 7. Since 7 is greater than 5, increase the hundredths digit to 7 and remove the remaining digits. The answer is 1,234.57.
The midpoint between the bounding multiples is 1,234.565. The original number lies above that midpoint, so the upper multiple is nearer. Subtracting the original from the answer gives 1,234.57 − 1,234.5678 = +0.0022.
Place-value rounding at a glance
| Round to | Step | Result |
|---|---|---|
| Thousands | 1,000 | 1,000 |
| Hundreds | 100 | 1,200 |
| Tens | 10 | 1,230 |
| Ones | 1 | 1,235 |
| Tenths | 0.1 | 1,234.6 |
| HundredthsWorked example | 0.01 | 1,234.57 |
| Thousandths | 0.001 | 1,234.568 |
Comparing the nine rounding rules
The first five rules choose the nearest value and differ only at an exact tie. The final four always follow their stated direction when a change is needed. In the table, all inputs are rounded to an integer.
| Rule | 2.5 | 3.5 | −2.5 |
|---|---|---|---|
| Standard: halves away from zero | 3 | 4 | −3 |
| Halves to even: banker's rounding | 2 | 4 | −2 |
| Halves up: ties toward +∞ | 3 | 4 | −2 |
| Halves down: ties toward −∞ | 2 | 3 | −3 |
| Halves toward zero | 2 | 3 | −2 |
| Always up: ceiling, toward +∞ | 3 | 4 | −2 |
| Always down: floor, toward −∞ | 2 | 3 | −3 |
| Toward zero: truncate | 2 | 3 | −2 |
| Away from zero | 3 | 4 | −3 |
NIST's Guide to the SI gives the half-to-even rule for exact ties. A discarded 5 followed only by zeros leaves an even retained digit unchanged and increases an odd retained digit. For custom multiples, evenness applies to the integer after division by the step: 12.5 rounded to the nearest 5 under half-to-even is 10, because 12.5 ÷ 5 = 2.5 rounds to 2.
Tips for getting the intended result
- 1Allow a carry through the digits. Rounding can change digits to the left of your target. For example, 9.996 to two decimal places becomes 10.00 because the increase carries through both decimal digits.
- 2Round directly from the original. Under Standard, 2.449 rounded directly to one decimal place is 2.4. Rounding first to 2.45 and then to one decimal place gives 2.5, a double-rounding difference.
- 3Use a direction only when the task requires it. Ceiling is useful for a positive count of containers when every item must fit. Floor is useful for the number of complete units that fit within a positive limit. Neither is a substitute for nearest-value rounding.
Rounding FAQ
How do you round to the nearest tenth?
Under Standard, look at the hundredths digit. If it is 5 or more, increase the tenths digit; otherwise, leave it unchanged, then remove the later digits. For example, 3.47 becomes 3.5 and 3.43 becomes 3.4.
Does 2.5 round to 2 or 3?
Standard gives 3 because exact ties go away from zero. Half-to-even gives 2 because 2 is the even neighboring integer. The selected rule determines the answer.
How do you round negative numbers?
Under Standard, round the magnitude and restore the minus sign: −2.6 becomes −3, −2.4 becomes −2 and −2.5 becomes −3. Ceiling instead takes −2.5 to −2, since −2 is numerically larger.
What is the difference between rounding and truncating?
Truncating removes unwanted digits without considering their size, so 2.99 truncated to an integer is 2, while Standard rounding gives 3. Truncation moves toward zero: −2.99 truncates to −2, not −3.
Should I keep zeros after the decimal point?
Keep them when you need to show a specified number of decimal places. For example, 3.999 to two decimal places is written 4.00. Those zeros show the requested rounding place; they do not establish the accuracy of an underlying measurement.
Are decimal places the same as significant figures?
No. Decimal places count positions after the decimal point, while significant figures begin with the first nonzero digit. For example, 0.004567 rounded to two decimal places is 0.00, but rounded to two significant figures it is 0.0046.
Why do different programs sometimes give different rounded answers?
They may use different tie rules, or store decimal inputs as binary approximations that affect midpoint comparisons. As an exact decimal, 1.005 rounded to two decimal places under Standard is 1.01. Check both the rounding rule and the program's number representation when comparing results.
Sources
- 1Prealgebra 2e, 1.1 Introduction to Whole Numbers · OpenStax, Rice University · Accessed November 2026
- 2Prealgebra 2e, 5.1 Decimals · OpenStax, Rice University · Accessed November 2026
- 3NIST Guide to the SI, Appendix B: Conversion Factors, Section B.7.1 Rounding numbers · National Institute of Standards and Technology · Accessed November 2026
- 4RoundingMode, Java SE 21 API documentation · Oracle · Accessed November 2026
- 5decimal: Decimal fixed-point and floating-point arithmetic · Python Software Foundation · Accessed November 2026