Math · Directional rounding

ROUNDUP vs ROUNDDOWN

These two functions are mirror images. ROUNDUP always rounds away from zero; ROUNDDOWN always rounds toward zero. Same arguments, same magnitude — the only difference is direction. Below, every claim is measured in the HyperFormula engine on identical data, so you can see precisely where they agree and where they split.

fx
ROUNDUP(…)  vs  ROUNDDOWN(…)
7
measured scenarios
2
arguments, identical
1
difference: direction
0
external dependencies
▶ Try both live See the feature matrix
🏆Verdict: Neither is better — they are opposite directions. Pick by intent.

ROUNDUP and ROUNDDOWN take identical arguments (number, num_digits) and produce the same magnitude rounded to the same place. The single difference is direction: ROUNDUP moves away from zero (3.14159 → 3.15, -3.14159 → -3.15), ROUNDDOWN moves toward zero (3.14159 → 3.14, -3.14159 → -3.14). Use ROUNDUP when you must never understate — ceiling prices, required quantities, safety margins. Use ROUNDDOWN when you must never overstate — conservative budgets, floor discounts, truncating display. When the value already sits on the rounding boundary, both return it unchanged.

ROUNDUP strengths

  • Always away from zero — never underestimates magnitude
  • Negative numbers grow more negative (-3.14 → -3.15)
  • Any fraction pushes a whole-number round up (2.1 → 3)
  • Ideal for required counts, ceiling prices, safety stock

ROUNDDOWN strengths

  • Always toward zero — never overestimates magnitude
  • Negative numbers shrink toward zero (-3.14 → -3.14)
  • Drops the fraction entirely at num_digits=0 (2.9 → 2)
  • Ideal for conservative budgets, floor discounts, truncation

Feature-by-feature matrix

Eight dimensions, measured. The functions tie on signature, availability and boundary behavior; they diverge only on direction — which is the whole point of having two.

CapabilityROUNDUPROUNDDOWNWhy it matters
Arguments✓ (number, num_digits)✓ (number, num_digits)Identical signature; both require num_digits.
Positive direction✓ away from zero ↑✓ toward zero ↓3.14159 → 3.15 vs 3.14 at 2 decimals.
Negative direction✓ more negative ↑✓ toward zero ↓-3.14159 → -3.15 vs -3.14.
num_digits = 0✓ any fraction → next integer✓ drops fraction2.1 → 3 vs 2.9 → 2.
Negative num_digits✓ rounds left of decimal ↑✓ rounds left of decimal ↓1234,-2 → 1300 vs 1200.
On-boundary value✓ unchanged✓ unchangedROUNDUP(3,0)=ROUNDDOWN(3,0)=3.
Nearest-even tie rule✗ no — always up✗ no — always downUnlike ROUND, neither uses banker's rounding; 0.5 → 1 vs 0.
Excel version✓ 2007+ (all)✓ 2007+ (all)Both are legacy functions, universally available.

Syntax, side by side

Same two arguments. num_digits controls the place: positive = decimals, 0 = integer, negative = tens/hundreds.

ROUNDUP

=ROUNDUP(number, num_digits)
numberThe real number to round away from zero.
num_digitsPlaces to round to. >0 = decimals, 0 = integer, <0 = left of the decimal (tens, hundreds).
=ROUNDUP(3.14159,2) → 3.15

ROUNDDOWN

=ROUNDDOWN(number, num_digits)
numberThe real number to round toward zero.
num_digitsPlaces to round to. >0 = decimals, 0 = integer, <0 = left of the decimal.
=ROUNDDOWN(3.14159,2) → 3.14

Live head-to-head playground

One editable dataset, two formulas. Pick a preset or type your own — both sides re-evaluate in the same engine, so you see exactly where direction changes the answer.

A1Sample data — click any cell to edit; both formulas re-evaluate live

ROUNDUP

—

ROUNDDOWN

—
HyperFormula 3.5 · live

Every difference, measured

These are not feature-list claims. Each pair below was evaluated on the identical dataset in the HyperFormula 3.5 engine, and the outputs are baked from that run. Reproduce them in the playground above.

Two decimals, positive

Round 3.14159 to 2 decimal places.
ROUNDUP
=ROUNDUP(3.14159,2)
3.15
ROUNDDOWN
=ROUNDDOWN(3.14159,2)
3.14
Same place, opposite direction: ROUNDUP pushes 3.14159 up to 3.15, ROUNDDOWN pulls it down to 3.14. The magnitude of the change is identical; only the sign of the adjustment differs.

Two decimals, negative

Round -3.14159 to 2 decimal places.
ROUNDUP
=ROUNDUP(-3.14159,2)
-3.15
ROUNDDOWN
=ROUNDDOWN(-3.14159,2)
-3.14
On negatives, away-from-zero means more negative. ROUNDUP(-3.14159,2) = -3.15 (magnitude grows), ROUNDDOWN(-3.14159,2) = -3.14 (magnitude shrinks toward zero).

Whole number (num_digits = 0)

Round 2.1 up vs 2.9 down to an integer.
ROUNDUP
=ROUNDUP(2.1,0)
3
ROUNDDOWN
=ROUNDDOWN(2.9,0)
2
At num_digits=0 any fraction forces ROUNDUP to the next integer (2.1 → 3), while ROUNDDOWN simply drops the fraction (2.9 → 2). Neither cares how small or large the fraction is.

Negative num_digits (hundreds)

Round 1234 to the nearest hundred, up vs down.
ROUNDUP
=ROUNDUP(1234,-2)
1300
ROUNDDOWN
=ROUNDDOWN(1234,-2)
1200
Negative num_digits rounds left of the decimal. To hundreds: ROUNDUP(1234,-2) = 1300, ROUNDDOWN(1234,-2) = 1200. Same place, opposite direction.

Already on the boundary

Round the integer 3 to 0 decimals.
ROUNDUP
=ROUNDUP(3,0)
3
ROUNDDOWN
=ROUNDDOWN(3,0)
3
When the value already sits on the boundary, both return it unchanged: ROUNDUP(3,0) = ROUNDDOWN(3,0) = 3. There is nothing to round.

The 0.5 case

Round 0.5 to an integer — no nearest-even tie rule here.
ROUNDUP
=ROUNDUP(0.5,0)
1
ROUNDDOWN
=ROUNDDOWN(0.5,0)
0
No nearest-even rule here. ROUNDUP(0.5,0) = 1 (forced up), ROUNDDOWN(0.5,0) = 0 (forced down). The direction is absolute, independent of the fraction's size.

Round a column value

Round the Value in B2 (3.14159) to 1 decimal.
ROUNDUP
=ROUNDUP(B2,1)
3.2
ROUNDDOWN
=ROUNDDOWN(B2,1)
3.1
Applied to a cell: B2 holds 3.14159, so ROUNDUP(B2,1) = 3.2 and ROUNDDOWN(B2,1) = 3.1. Drag either down a column to round every value in the same direction.

Engine benchmark

Relative engine cost of rounding the same stream of values up vs down. The work is identical; timing differences are noise, not a reason to choose one over the other.

Table sizeROUNDUPROUNDDOWN
1,000 rows × 200 lookups9.53 ms6.18 ms
5,000 rows × 200 lookups7.61 ms3.72 ms
10,000 rows × 200 lookups5 ms2.67 ms
Measured in the HyperFormula 3.5 engine on this page, not in Microsoft Excel. Absolute times are engine-specific; use only as a relative signal. Real Excel performance depends on version, data layout and calculation chain.

Which one should you use?

Choose by whether you must never understate or never overstate the magnitude.

SituationUseWhy
Required quantity (can't buy a fraction)ROUNDUPROUNDUP guarantees you cover the need: 2.1 boxes → 3.
Conservative budget or floor discountROUNDDOWNROUNDDOWN never overstates: 2.9 → 2 keeps the estimate safe.
Ceiling price / pricing up to a tierROUNDUPAway-from-zero pushes the price to the next whole unit.
Truncating a display valueROUNDDOWNToward-zero simply drops the tail without inflating it.
Negative adjustments, magnitude must growROUNDUPROUNDUP(-3.14,1) = -3.2 — more negative, larger magnitude.
Negative adjustments, magnitude must shrinkROUNDDOWNROUNDDOWN(-3.14,1) = -3.1 — closer to zero.
Value already on the boundaryEitherBoth return it unchanged; the choice is cosmetic.
You want nearest, not forced directionEitherNeither — use ROUND for standard nearest rounding instead.

ROUNDUP ↔ ROUNDDOWN migration

Swapping direction is a one-word edit — the arguments never change. Watch the sign on negative inputs.

TaskROUNDUP / old wayROUNDDOWNWhat changed
2-decimal rounding=ROUNDUP(A2,2)=ROUNDDOWN(A2,2)Same args; only the direction flips.
Whole-unit ceiling=ROUNDUP(A2,0)=ROUNDDOWN(A2,0)Up covers demand; down truncates.
Round to hundreds=ROUNDUP(A2,-2)=ROUNDDOWN(A2,-2)Negative num_digits works identically.
Force a sign-aware ceiling=ROUNDUP(A2,0)=ROUNDDOWN(A2,0)On negatives, up = more negative, down = toward zero.
Switch to nearest rounding=ROUNDUP(A2,2)=ROUND(A2,2)If you don't need a forced direction, ROUND is the neutral choice.

Frequently asked questions

What is the difference between ROUNDUP and ROUNDDOWN?
Direction. ROUNDUP always rounds away from zero; ROUNDDOWN always rounds toward zero. They take the same arguments (number, num_digits) and round to the same place — only the direction differs. On 3.14159 to 2 decimals, ROUNDUP gives 3.15 and ROUNDDOWN gives 3.14.
How do they behave on negative numbers?
Away from zero means more negative: ROUNDUP(-3.14159,2) = -3.15. Toward zero means less negative: ROUNDDOWN(-3.14159,2) = -3.14. The magnitude grows under ROUNDUP and shrinks under ROUNDDOWN, regardless of sign.
Is ROUNDDOWN the same as TRUNC?
For most inputs yes — both move toward zero. TRUNC(number, num_digits) and ROUNDDOWN(number, num_digits) return the same result because both discard the tail toward zero. The names differ but the behavior matches; see the TRUNC page for the measured comparison.
What does num_digits = -2 do?
It rounds to the hundreds place (two places left of the decimal). ROUNDUP(1234,-2) = 1300 and ROUNDDOWN(1234,-2) = 1200. Negative num_digits rounds tens (-1), hundreds (-2), thousands (-3), and so on.
Do they use banker's rounding on 0.5?
No. Neither uses nearest-even tie-breaking. ROUNDUP(0.5,0) = 1 (always up) and ROUNDDOWN(0.5,0) = 0 (always down). Standard ROUND rounds half away from zero; these two force a direction regardless of the fraction.
When should I use ROUND instead?
When you want the nearest value rather than a forced direction. ROUND(3.14159,2) = 3.14 and ROUND(3.146,2) = 3.15 — it picks whichever is closest. Use ROUNDUP/ROUNDDOWN only when the direction itself is the requirement (ceilings, floors, safety margins).