Significant Figures: The Rules, the Ambiguous Cases, and How to Round Correctly

0.004560 has four significant figures, 105 has three, and 1200 is genuinely ambiguous, and once you see the four rules applied digit by digit, none of it is mysterious. Significant figures are just a way of saying how much of a number you actually know. Count and round any number, with each digit color-coded by rule, using the significant figures calculator; here’s the reasoning behind it.

The four rules

Every significant-figures question reduces to four rules about which digits carry information versus which are just holding the decimal point in place:

Rule	Example	Sig figs
Non-zero digits always count	42	2
Zeros between non-zero digits count	105, 4008	3, 4
Leading zeros never count	0.0056	2
Trailing zeros count only with a decimal point	100.0 vs 1200	4 vs ambiguous

Walk 0.004560 through them: the 0.00 at the front are leading zeros (rule 3, skip), counting begins at 4, and 4, 5, 6, 0 all count, the final 0 is a trailing zero after a decimal point, so rule 4 makes it significant. Four significant figures. The calculator colours exactly these digits so the count is never a guess.

The trailing-zero trap (and why scientific notation exists)

The hardest rule is trailing zeros, and it’s worth dwelling on because it’s where nearly all confusion lives. Compare:

• 100.0, four sig figs. The decimal point signals that every digit, zeros included, was measured and matters.
• 1200, ambiguous. Was it measured to the nearest 1, 10, or 100? A bare integer can’t say.

This ambiguity is the entire reason scientific notation is used in science: it forces you to state your precision. 1.2 × 10³ is unambiguously two sig figs; 1.20 × 10³ is three; 1.200 × 10³ is four. If you ever need to communicate that a trailing zero is meaningful, write it in scientific notation, that’s what the format is for. (The sig-fig calculator flags bare-integer trailing zeros as ambiguous rather than pretending to know.)

Exact numbers have unlimited significant figures

One category sits outside the four rules entirely: exact numbers. These are quantities that carry no measurement uncertainty at all, so they never limit the precision of a result.

• Counted objects. If you count 24 students in a room, that 24 is exact. There is no “24.0 give or take.” It behaves as if it had infinitely many significant figures.
• Defined conversion factors. The relationships 1 m = 100 cm, 1 in = 2.54 cm, and 1 min = 60 s are definitions, not measurements. The 100, the 2.54, and the 60 are exact.
• Integers in formulas. The 2 in 2πr or the 4 in the area of a triangle formula is a pure mathematical constant, not a measured value.

The practical upshot: when you divide a measured mass by a counted number of items, the count does not drag your answer down to fewer significant figures. Only the measured inputs set the limit.

Rounding to N significant figures

Rounding to significant figures works like ordinary rounding, but you count from the first significant digit rather than the decimal point:

1. Identify the first N significant digits.
2. Look at the next digit: 5 or more rounds up, less than 5 stays.
3. Replace the rest with zeros (or drop them, past the decimal point).

Examples:

Number	To 2 s.f.	To 3 s.f.
0.004560	0.0046	0.00456
12345	12000	12300
9.99	10	9.99

That last row is the one people get wrong: rounding 9.99 to two significant figures rounds the second 9 up, which carries, 9.9 → 10, not 9.9. Exact digit-string rounding (which the tool uses) handles the carry correctly where naive float rounding can drift.

There is one subtlety when the digit you drop is exactly 5 with nothing after it. Most schools teach round half up (always round the 5 up), which is simple and predictable. Some scientific and computing contexts prefer round half to even (also called banker’s rounding): round to make the kept digit even, so 2.5 and 3.5 both round to a nearest-even result. Round-half-to-even avoids a slight upward bias when you round many values, which is why it is the IEEE-754 default for floating point. For everyday work, round half up is fine, just pick one convention and apply it consistently.

Why the rules exist

Significant figures aren’t arbitrary bookkeeping. They encode measurement precision. A ruler reading of 2.0 cm claims you know the length to a tenth of a centimetre; 2 cm claims only to the nearest centimetre. When you calculate with measured values, the result can’t be more precise than its least-precise input: multiply a 3-sig-fig measurement by a 2-sig-fig one and the answer gets 2 sig figs, because that’s all you actually know. This is why chemistry, physics and engineering carry sig figs through every step and round at the end.

Two quick operational rules follow:

• Multiplication/division: the result keeps the fewest significant figures of any input.
• Addition/subtraction: the result keeps the fewest decimal places of any input (a different rule, it’s about decimal position, not sig-fig count).

A worked calculation

Say you measure a density: mass 4.50 g (three sig figs) divided by volume 1.2 mL (two sig figs). The raw quotient is 3.75 g/mL, but the multiplication/division rule caps the answer at the fewest input sig figs, two, so you report 3.8 g/mL. Carrying the full 3.75 through any further steps and rounding only at the very end avoids compounding rounding error; you round the final reported value, not every intermediate one.

Contrast that with addition. Add lengths 12.11 cm (two decimal places) and 1.1 cm (one decimal place). The sum is 13.21 cm, but the addition rule keeps the fewest decimal places, one, so you report 13.2 cm. Notice the two rules can give different answers on the same digits, which is exactly why they are stated separately.

Common mistakes at a glance

Mistake	Wrong	Right	Why
Counting leading zeros	0.0056 has 4 s.f.	2 s.f.	Leading zeros only place the decimal
Dropping a significant trailing zero	100.0 has 3 s.f.	4 s.f.	The decimal point makes them count
Assuming 1200 is exact	1200 has 4 s.f.	ambiguous	A bare integer can’t say
Missing the rounding carry	9.99 → 9.9 at 2 s.f.	10	The carry ripples left
Mixing the two operation rules	using sig-fig count for a sum	use decimal places	Addition tracks position, not count

Quick summary

Significant figures answer “how much of this number do I actually know?” Non-zero digits and trapped zeros count; leading zeros never do; trailing zeros count only with a decimal point, which is why 1200 is ambiguous and scientific notation exists to fix it. Round from the first significant digit and mind the carry (9.99 → 10 at two sig figs). Count, round, and see every digit classified by rule with the significant figures calculator, and reach for scientific notation whenever a trailing zero needs to speak.

Sources: NIST, Uncertainty of measurement results · Wikipedia, significant figures · standard analytical-chemistry conventions (Harris, Quantitative Chemical Analysis).