ROUND()
Rounds a number to a specified number of decimal places.
ROUND(number, num_digits)ROUND rounds a number to the number of digits you specify. Positive num_digits rounds to that many decimal places: 2 gives two decimal places. Zero rounds to the nearest integer. Negative num_digits rounds to the left of the decimal: -1 rounds to the nearest 10, -2 to the nearest 100.
ROUND uses standard rounding — 0.5 rounds up. For always-round-up use ROUNDUP; for always-round-down use ROUNDDOWN or INT.
Rounding in a formula is different from formatting a cell with fewer decimal places. Formatting changes the display; ROUND changes the actual stored value.
ROUND works in decimal places, and a negative digit count rounds to tens, hundreds or thousands — useful for presenting large figures. For rounding to a multiple that is not a power of ten, such as the nearest 5p or the nearest quarter hour, MROUND is the right function, with CEILING and FLOOR for forcing the direction. The practical warning is about money: rounding each line and summing gives a different total from summing and rounding once, so decide where the rounding happens and apply it consistently across the model.
Arguments
- number
- The number to round.
- num_digits
- The number of digits to round to. Positive = decimal places; 0 = nearest integer; negative = tens, hundreds, etc.
Examples

=ROUND(3.14159, 2)Rounds to 2 decimal places.
=ROUND(1456, -2)Rounds to the nearest 100.
=MROUND(A2, 0.05)ROUND works in decimal places; MROUND rounds to a multiple, which is what pricing to the nearest 5p or 25 cents actually needs.
Notes and gotchas
- Excel's ROUND uses "round half away from zero": 2.5 goes to 3 and -2.5 goes to -3. This differs from the banker's rounding used by many other systems, which is why Excel totals sometimes disagree with a database by a penny.
- ROUND changes the stored value. Number formatting changes only the display. Use formatting when you want the underlying precision kept.
- Rounding each row then summing does not give the same answer as summing then rounding. Decide which one the business wants and apply it consistently.
- A negative
num_digitsrounds to tens, hundreds, thousands:ROUND(1234, -3)is 1000. - Comparing floating-point results needs rounding:
ROUND(a-b, 2)=0rather thana=b, or a control total will show a spurious failure.
Common errors
| You see | Because | Fix |
|---|---|---|
| #VALUE! | Text passed as the number or the digit count. | Verify both arguments are numeric. |
| Rounded column does not foot to the rounded total | Rounding was applied per row, and the residuals accumulate. | Decide whether the total or the rows are authoritative, and derive the other from it. |
ROUND tricks, tips, and workbook hacks
ROUND permanently alters the stored numeric value of a cell to the decimal place you specify, whereas Format Cells (Ctrl+1) only changes how the number is painted on screen while keeping full floating-point precision underneath. That distinction is the #1 cause of financial statements whose column appears to add up to $100.01 instead of $100.00.
Whenever a calculated line item represents actual currency billed on an invoice, posted to a general ledger, or paid via payroll, wrap the line calculation in ROUND(..., 2) so the stored pennies match the displayed pennies.
Quick ROUND checks before trusting the result
- Wrap invoice line extensions (Qty * UnitPrice * (1 - Discount)) in ROUND(..., 2) so the bottom-line SUM() always equals the sum of the visible pennies.
- Use negative num_digits (-3 for thousands, -6 for millions) when rounding high-level budget estimates.
- Leave intermediate rate calculations (like daily interest rates or unit cost weights) unrounded, and apply ROUND() only at the transaction or reporting boundary.
| ROUND pattern | ROUND formula | Best use | Watch out |
|---|---|---|---|
| Penny-exact invoice line total | =ROUND(B2*C2*(1-D2),2) | Ensuring that discounted line items store exact 2-decimal currency amounts before summing. | Without ROUND(..., 2), three lines ending in $0.004 display as .00 but add up to an extra $0.01 in SUM(). |
| Eliminate IEEE 754 floating-point subtraction dust | =ROUND(A2-B2,6) | Preventing tests like =(A2-B2)=0 from failing when binary arithmetic leaves 0.00000000000000022. | Critical before comparing calculated decimals inside IF() or XLOOKUP(). |
| Round to the nearest $500 or $1,000 | =ROUND(B2,-3) | Rounding annual salary bands or capital expenditure requests to whole thousands. | For non-power-of-10 multiples like nearest $25 or $50, use MROUND(B2, 50) instead. |
| ROUND mistake | What happens in ROUND | Fix |
|---|---|---|
| Relying on the Decrease Decimal ribbon button for accounting totals | Three rows showing $10.33, $10.33, and $10.33 sum to $31.00 instead of $30.99 on a printed client invoice. | Wrap each row's formula in =ROUND(..., 2) so the underlying stored value matches what is printed. |
| Rounding intermediate unit rates too early | Multiplying a prematurely rounded 2-decimal unit rate by 500,000 units introduces a $2,000 discrepancy. | Keep full precision on intermediate rates and apply ROUND(..., 2) to the final extended dollar amount. |
ROUND practice drills
- Enter =10/3 in three cells B2:B4, format them to 2 decimal places ($3.33), and check =SUM(B2:B4) ($10.00 vs $9.99). Now wrap B2:B4 in =ROUND(10/3, 2) and compare.



