How to Round Numbers in Excel: ROUND, UP, DOWN, MROUND

Rounding numbers in Excel sounds boring until you've actually messed it up. A finance manager who showed $1,000,000 as $999,999.50 in a board pack. A teacher who marked an entire spreadsheet of grades as 79.5% when they should have been 80%. A pricing analyst whose retail prices ended in .493 because someone used the wrong function.

Excel gives you about seven different ways to round. Most people use two of them. The right one depends on what the number means.

This guide walks through every Excel rounding function — ROUND, ROUNDUP, ROUNDDOWN, MROUND, INT, FLOOR, and CEILING — with concrete examples for decimals, whole numbers, and negative digits. We'll cover when to use which. You'll see the exact syntax for each, the rookie traps that catch even experienced users, and a few tricks for rounding to the nearest $5, $25, or quarter-hour.

If you only remember one thing — the second argument matters more than the first. That's where almost every rounding mistake lives.

Start with the workhorse. The ROUND function takes two arguments — the number you want to round, and the number of decimal places. Syntax looks like this: =ROUND(number, num_digits). That's it.

Round 3.14159 to two decimals? =ROUND(3.14159, 2) returns 3.14. Round it to zero decimals? =ROUND(3.14159, 0) returns 3. So far, so simple.

Here's where it gets interesting. The num_digits argument can be negative. Set it to -1, and Excel rounds to the nearest 10. Set it to -2, nearest 100. Set it to -3, nearest 1,000. So =ROUND(1547, -2) returns 1500, while =ROUND(1547, -3) returns 2000. That single feature solves about 80% of real-world rounding problems — call it "round to a round number" and you'll never forget it.

The function works on text values too, if Excel can coerce them to numbers. =ROUND("3.14159", 2) returns 3.14 just like the numeric version. If the text isn't a valid number — say "abc" — you get #VALUE!. Useful when pulling rounded values out of imported CSV files where numbers came in as text.

What about ties? When the digit you're rounding is exactly 5, Excel rounds away from zero. So =ROUND(2.5, 0) returns 3, and =ROUND(-2.5, 0) returns -3. This is sometimes called "arithmetic rounding" or "round half up" for positives.

It's different from "banker's rounding" (round half to even), which spreadsheet purists and some statisticians prefer because it removes bias from large datasets. Excel doesn't have a built-in banker's-rounding function — you'd have to build one with MROUND plus IF logic, or use Power Query's Number.Round with rounding mode 2.

For 99% of everyday work, the default ROUND behavior is what you want. Just know that it exists, so when an auditor asks "did you bank-round these averages?" you can give an honest answer.

One more rounding quirk people don't expect — IEEE 754 floating-point precision. Excel stores numbers as 64-bit doubles, which means very long decimals can't be represented exactly. Type 0.1 + 0.2 in a cell and Excel internally has 0.30000000000000004. Most of the time you'll never notice. Once in a while it produces a "rounded value differs by 0.01" bug that looks impossible. Wrapping arithmetic in ROUND fixes it cleanly.

Next up — ROUNDUP and ROUNDDOWN. These force the direction regardless of the digit. =ROUNDUP(3.001, 0) returns 4. Even though .001 is barely above three, ROUNDUP doesn't care. It always rounds away from zero.

ROUNDDOWN does the opposite. =ROUNDDOWN(3.999, 0) returns 3. Always toward zero. The negative-number behavior catches people. =ROUNDUP(-3.1, 0) returns -4, because "away from zero" means more negative. =ROUNDDOWN(-3.9, 0) returns -3, because "toward zero" means less negative. Backwards from what most users expect on first read.

When do you use these instead of ROUND? Billing scenarios are the classic example. A telecom company rounds usage up so a 30.1-minute call gets billed as 31 minutes. A web hosting bill might round storage up to the next gigabyte. Anything where the business case demands "always over, never under" — that's ROUNDUP. And anything where you want to under-promise — like estimating delivery time slots — that's ROUNDDOWN.

The MROUND function works differently from the others. Instead of rounding to a number of decimal places, it rounds to the nearest multiple of a number you specify. Syntax: =MROUND(number, multiple).

This is the function for rounding to the nearest five, ten, twenty-five, or any custom step. Want prices that always end in .99? =MROUND(price, 0.99) won't do that exactly — but =ROUND(price, 0) - 0.01 will. For nickel rounding (common in cash-only countries that dropped the penny), =MROUND(15.43, 0.05) returns 15.45.

The classic real-world MROUND use is time. To round a meeting duration to the nearest 15 minutes — =MROUND(A1, "0:15") where A1 holds the duration. That single formula saves payroll teams hours a week. It also handles negative numbers if the multiple sign matches: both must be positive or both negative, otherwise MROUND returns #NUM!. That quirk is documented but trips people up.

Now, the integer family. INT, FLOOR, CEILING, and TRUNC look similar but behave differently in important ways — especially with negatives.

INT(number) returns the integer part by rounding down toward negative infinity. So =INT(3.9) returns 3, and =INT(-3.9) returns -4. That last one matters. INT doesn't strip the decimal — it actively rounds down. For -3.9, "down" means -4, not -3.

TRUNC(number) chops the decimal off. =TRUNC(3.9) returns 3, and =TRUNC(-3.9) returns -3. So for positive numbers, INT and TRUNC behave identically. For negatives, they go opposite directions. If you want to throw the decimal away regardless of sign — use TRUNC. If you want true mathematical floor behavior — use INT.

This distinction broke a payroll system at a company I worked with. They were using INT for negative adjustments and ending up with $1 extra deductions every time. Switched to TRUNC, problem gone.

FLOOR and CEILING are MROUND's directional cousins. =FLOOR(number, significance) rounds down to the nearest multiple. =CEILING(number, significance) rounds up. Both take the same two-argument structure.

So =FLOOR(47, 10) returns 40. =CEILING(47, 10) returns 50. =FLOOR(47, 25) returns 25. =CEILING(47, 25) returns 50. They're MROUND with a forced direction.

In Excel 2013 and later, FLOOR.MATH and CEILING.MATH replaced the older versions because they handle negatives more predictably. The new functions accept an optional third argument — mode — which controls negative-number behavior. =CEILING.MATH(-4.7, 1, 1) returns -5 (away from zero), while =CEILING.MATH(-4.7, 1, 0) returns -4 (toward zero, the default). For new spreadsheets, default to the .MATH versions. They're cleaner.

Let's compare them side by side with one example — the number 4.5. Watch what each function does at zero decimal places.

ROUND(4.5, 0) returns 5 (round half up). ROUNDUP(4.5, 0) returns 5. ROUNDDOWN(4.5, 0) returns 4. INT(4.5) returns 4. TRUNC(4.5) returns 4. MROUND(4.5, 1) returns 5 (rounds to nearest 1). FLOOR(4.5, 1) returns 4. CEILING(4.5, 1) returns 5.

Now -4.5. ROUND(-4.5, 0) returns -5. ROUNDUP(-4.5, 0) returns -5 (away from zero). ROUNDDOWN(-4.5, 0) returns -4 (toward zero). INT(-4.5) returns -5 (down toward minus infinity). TRUNC(-4.5) returns -4 (chop decimal). The split between INT and TRUNC for negatives is the single most common rounding bug in financial models.

Memorize that table once and rounding mistakes mostly disappear. Print it. Tape it to your monitor. The first time you have to debug a $0.02 discrepancy in a 50,000-row revenue report, you'll thank yourself for the cheat sheet. Excel won't warn you. The wrong function returns a number that looks right but isn't.

A note on cell formatting versus actual rounding. Right-clicking a cell, selecting Format Cells, and reducing decimal places does not change the underlying value. The cell still holds 3.14159 even if it displays 3.14. Any formula referencing that cell uses the full precision number.

This is the cause of countless "Excel can't add" complaints. A user sees 100 + 200 = 301, can't explain why, and blames Microsoft. The reality is one of those cells contains 100.49 displayed as 100, and the other contains 200.5 displayed as 200. Sum is 300.99, which Excel displays as 301. Math is right. Formatting was hiding the decimals.

The fix is either using ROUND inside the formula — =ROUND(A1, 0) + ROUND(B1, 0) — or turning on Excel's "Set precision as displayed" option in File → Options → Advanced. Be careful with that second option. It permanently rounds the actual stored values to whatever precision the cells are formatted to. Once you save, the original precision is gone.

What about rounding times and dates? Excel stores both as serial numbers (dates as days since 1900, times as fractions of a day). So all the same rounding functions work, you just need to think in decimal-of-day units.

Want to round a timestamp to the nearest 15 minutes? Fifteen minutes is 1/96 of a day, so =MROUND(A1, 1/96) works, or more readably =MROUND(A1, TIME(0,15,0)). To the nearest hour — =MROUND(A1, 1/24) or =MROUND(A1, TIME(1,0,0)).

For payroll specifically, the "round to nearest quarter hour" rule (sometimes called the 7-minute rule under FLSA) means times between :00-:07 round down to :00, :08-:22 round to :15, and so on. =MROUND(clock_in, "0:15") does exactly that. Just confirm with your HR or compliance team that arithmetic rounding (versus rounding only in the employee's favor) is permitted in your jurisdiction. California, for example, recently restricted neutral time-rounding.

Currency rounding deserves its own paragraph because of how often it goes wrong. Most currencies use two decimal places, so most users default to =ROUND(value, 2). That's correct for USD, EUR, GBP, and most major currencies.

But some currencies don't. The Japanese yen (JPY) has no minor unit — round to zero decimals. The Kuwaiti dinar (KWD) and Bahraini dinar (BHD) use three decimals. If your spreadsheet handles multi-currency data, hard-coding 2 in your ROUND function will silently corrupt JPY values by 0.0049 every row, which compounds across thousands of rows into real money.

A safer pattern — store the decimal count for each currency in a lookup table and use VLOOKUP or XLOOKUP to feed the ROUND second argument. Something like =ROUND(value, VLOOKUP(currency_code, decimals_table, 2, FALSE)). Tedious to set up once, then bulletproof forever.

Another scenario worth covering — rounding inside a SUMPRODUCT or array formula. If you have unrounded values and want the sum to use rounded contributions, you need to push ROUND inside the array.

So instead of =ROUND(SUMPRODUCT(prices, qtys), 2) — which sums first, then rounds the total — you want =SUMPRODUCT(ROUND(prices*qtys, 2)), which rounds each line then sums. The difference can be material on long lists. A 50-row invoice with line items ranging from $0.001 to $0.009 will produce a different total depending on which method you use.

The "round each line, then sum" approach is what most accounting standards require for VAT, GST, and sales tax calculations. Check your local tax authority's specific guidance. Most published tax rounding rules can be expressed as a combination of ROUND and SUMPRODUCT or SUMIF.

One last function worth mentioning — EVEN and ODD. =EVEN(7.3) returns 8 (rounds up to the next even integer). =ODD(7.3) returns 9 (rounds up to the next odd integer). For negatives, both round away from zero, so =EVEN(-7.3) returns -8 and =ODD(-7.3) returns -9.

Niche functions, but they have their uses. Packaging sizes that have to be in multiples of two (printing print runs, shipping cartons that pair up). Sports brackets that need an even number of competitors. Engineering specs where odd-numbered fittings have different stocking patterns.

They round in only one direction — up, away from zero — which means they're not symmetric with the other ROUND functions. If you find yourself reaching for EVEN or ODD often, double-check that MROUND with a multiple of 2 isn't a better fit. MROUND can round to nearest, not just up.

Before wrapping, let's talk about why this matters at all. Rounding errors compound. Each individual rounding decision looks tiny — half a cent here, a tenth of a percentage point there. Across a million rows, those become discrepancies that auditors notice.

Financial reports, scientific datasets, billing systems, payroll runs — every one of those workflows has been broken at some point by inconsistent rounding. The fix is usually trivial once you find it. Finding it takes hours. The pattern is always the same. Someone used ROUND when they should have used MROUND, or used INT when they should have used TRUNC, or formatted the cell to two decimals and assumed Excel followed along internally.

The remedy is to standardize. Pick a rounding policy at the start of any model. Document which function gets used for what. Use named ranges or named formulas if your team is on Excel 365. When someone joins the team, hand them the rounding cheat sheet on day one. The hour you spend documenting saves a week of debugging downstream.

If you're studying for Excel certifications — check the Excel functions reference for full syntax tables, or our formulas cheat sheet for printable quick references covering all 460+ built-in functions.

One final tip — when you build a rounding-heavy model, add a hidden audit row that shows the raw unrounded value next to the displayed one. Format it gray. Hide it before sending the workbook out. If something looks wrong later, you have the original numbers one keystroke away. Future-you will be grateful.