Free Percentage Calculator

A Short History of "Per Cent"

The English word "percent" descends from Modern Latin per centum: literally "by the hundred", formed from the preposition per ("through, by means of") plus centum ("hundred"). The Oxford English Dictionary records the earliest English evidence in 1568, in a letter by the financier Sir Thomas Gresham, and "per cent" (two words) is dated to the 1560s. Until the early twentieth century it was often treated as an abbreviation and written as "per cent." with a closing period. The compound noun "percentage" follows much later, around 1789. The mathematical idea predates the word by many centuries: Augustus levied the centesima rerum venalium, a one-percent tax on goods sold at public auction that funded the Roman military retirement fund. The Romans did not write "1%" (their numerals had no symbol for it and their fractions were duodecimal) but the conceptual move of expressing a tax rate as a fixed fraction of one hundred is the same one a modern percent does.

The percent symbol itself has a much better-documented history. Before about 1425, no special glyph existed; merchants wrote out per cento or used variant abbreviations like "per 100" or "p cento". The pivotal step is captured in an anonymous Italian manuscript from 1425 where a scribe writes "pc" with a small loop adorning the "c" to represent the Italian ordinal ending -o. Over the next two centuries the pc-with-a-loop mutated. By around 1650, the "pc" had collapsed into something resembling a horizontal fraction bar and the per part dropped out entirely; a 1684 text shows it in a form recognisable to a modern reader. The diagonal-slash form of the modern % symbol is, as the historian D. E. Smith noted in 1925, "modern", late-nineteenth or early-twentieth century. The everyday symbol you tap on a keyboard is the residue of a five-hundred-year game of typographic telephone: an Italian merchant's scrawled abbreviation for per cento, gradually compacted into a slash flanked by two small circles.

Three Operations That Look Alike but Aren't

A percentage calculator that earns its name covers three operations that look superficially related but are not interchangeable:

• Percentage of a whole. "What is X% of Y?" computes X × Y / 100. The percent is the operator; the whole is the operand. 18% of 47 = 0.18 × 47 = 8.46. This is the operation people use to add a tip, take a discount or compute a tax line.
• What percentage one number is of another. "A is what percent of B?" computes A × 100 / B. Here the unknown is the percent itself; the two numerical inputs are the part and the whole. 8.46 is what percent of 47 → 8.46 × 100 / 47 = 18%. This is the operation people use to figure a grade out of a maximum, a market share out of a total or a vote share.
• Percentage change. "What is the percentage change from old to new?" computes (new − old) / |old| × 100. The denominator is the original value; the absolute value ensures the formula behaves sensibly when the starting figure is negative. The U.S. Bureau of Labor Statistics codifies the same definition for the consumer price index. From 50 to 60 is a 20% increase; from 60 to 50 is roughly a 16.7% decrease. Note the asymmetry, the same dollar change gives different percentages depending on which direction you go.

Three things conspire to make these operations hard to keep straight. The same English word "of" appears in both (a) and (b), so people parsing the sentence have to read the question structure to know which way to point the formula. Percentage change is non-symmetric: if a value goes up by 25% it must come down by 20% to return to its starting point, not 25%. And percentages of percentages do not add, a 10% rise followed by a 10% fall lands at 0.99 of the original, not back at 1.00. The empirical evidence backs up the everyday intuition that this stuff is hard: a study of 1,629 university students found that even those who had taken mathematics through Calculus II answered correctly on only about 75% of basic two-step percentage questions.

Percentage Points vs Percent, and Basis Points

A rate that moves from 5% to 6% has gone up by one percentage point. It has also gone up by twenty percent: because 1 is 20% of 5. Both descriptions are true and both refer to the same underlying change, but they sound very different and they answer different questions. The percentage point (often abbreviated "pp") is the unit for the absolute, additive change in a rate: 6 − 5 = 1 pp. The percent is the unit for the relative, multiplicative change in the rate, treating the starting rate as the base: (6 − 5) / 5 = 20%. In financial markets the unit-of-account problem is solved by using basis points. One basis point is one hundredth of a percentage point. A move from 4.25% to 4.50% is therefore 25 basis points (bps); 100 bps equals 1 percentage point. The Federal Reserve's Federal Open Market Committee uses this convention rigorously: a typical FOMC press release describes a rate decision as "lowering the target range for the federal funds rate by 1/4 percentage point," with market commentary then reiterating it as "25 basis points." Basis points remove the ambiguity entirely, because "100 basis points" can only mean 100/10,000 = 0.01 in absolute terms (never a relative change. The classic journalist trap is to write "the unemployment rate fell 1 percent" when the rate went from 4% to 3%) really a 1 percentage-point fall and a 25% relative fall. Politicians exploit the ambiguity in both directions: a small absolute change can be packaged as a large relative one ("a 25% drop in unemployment") and a large absolute change can be downplayed as small ("only 1 percent"). The rule: when reporting changes in things that are themselves rates (interest, unemployment, tax brackets, vote share), use percentage points for the additive change and reserve "percent" for the multiplicative one.

Compound Interest and the Rule of 72

Compound interest is the canonical application of percentages to time. If you have principal P invested at annual rate r (as a decimal), then after t years of annual compounding you have P(1 + r)^t. The exact doubling time is t = ln(2) / ln(1 + r). The Rule of 72 is a mental-arithmetic shortcut: at an interest rate of r percent per period, the doubling time in periods is approximately 72 / r. At 6%, money doubles in roughly 12 years; at 8%, in 9 years; at 1%, in 72 years. The earliest documented appearance of the rule is in Luca Pacioli's Summa de arithmetica, published in Venice in 1494: "tieni per regola 72, a mente, il quale sempre partirai per l'interesse": "keep as a rule the number 72 in mind, which you will always divide by the interest." Pacioli does not derive the rule, which suggests the heuristic was already in use among Italian merchants and that he was passing on practical wisdom rather than a discovery of his own. Old Babylonian clay tablets from around 2000 BCE pose problems equivalent to "how long does it take a principal to double at 20% compounded annually?", tablets AO 6770 and VAT 8528 in the Berlin museum work through such problems explicitly. The continuous-compounding limit is the next conceptual step: if you take P(1 + r/n)^nt and let n grow without bound, you get P · e^rt, where the constant e ≈ 2.71828 was first identified by Jacob Bernoulli in 1683 studying exactly this question. Leonhard Euler adopted the letter e for it in correspondence dated 1727 or 1728.

Markup vs Margin, The $30 Trap

Retail and wholesale trade run on two related but distinct percentages, and the difference matters for any user who is doing actual pricing math. Markup is the profit expressed as a percentage of cost: (Selling Price − Cost) / Cost × 100. Margin (more precisely, gross margin) is the profit expressed as a percentage of selling price: (Selling Price − Cost) / Selling Price × 100. The numerator is the same dollar amount in both formulas (the gross profit) but the denominator differs. As a result, markup is always larger than margin for the same item. Sell something for $100 at a cost of $70: profit is $30, markup is 30/70 = 42.9%, margin is 30/100 = 30%. The conversion formulas, with both quantities as decimals, are clean: Margin = Markup / (1 + Markup); Markup = Margin / (1 − Margin). So a 50% markup equals a 33.3% margin; a 50% margin equals a 100% markup. Why two conventions? Buyers and category managers in retail tend to think in markup because cost is what they pay the supplier and the markup is what they decide to add. Accountants and CFOs think in margin because selling price is what hits the income statement as revenue. The classic small-business mistake is to set a price by adding what feels like a healthy markup and then assume the resulting margin is the same number. A product purchased for $50 and "marked up 50%" sells at $75, but the margin on that sale is only 33.3%. If a target gross margin of 50% is what the business actually needs to cover overhead, the markup must be 100%, not 50%.

Sales Tax, VAT, and the Reverse-Percentage Trap

The tax line on a receipt is the most universal percentage calculation in everyday life. The United States uses sales tax, levied only at the final point of sale to a consumer, with rates set independently by states, counties and municipalities, there is no federal sales tax. Five states have no statewide sales tax (the NOMAD acronym: New Hampshire, Oregon, Montana, Alaska, Delaware). California has the highest statewide rate at 7.25%, with combined local rates pushing as high as 10.75% in some jurisdictions; the population-weighted national average combined rate is 7.53%. In the US tradition prices are usually displayed exclusive of sales tax, the shelf tag shows $9.99, the receipt adds tax on top. Most of the rest of the world uses value-added tax (VAT) or its near-synonym goods and services tax (GST). VAT was invented by the French civil servant Maurice Lauré, who first deployed it in France's Côte d'Ivoire colony on 10 April 1954 before introducing it domestically in 1958. The mechanism differs structurally: each business in the supply chain charges VAT on its sales and reclaims VAT paid on its inputs, so the tax accrues to the government in slices at every stage rather than landing entirely at the final retail register. In jurisdictions like the UK and the EU, prices to consumers are usually displayed inclusive of VAT. The UK standard rate is 20%; Hungary tops European standard rates at 27%; the typical EU-member rate sits in the 19-23% range. Canada uses a federal 5% GST plus provincial sales taxes; Australia uses 10% GST.

The "reverse percentage" problem is asking: given a tax-inclusive total, what was the price before tax? The intuitively wrong answer is to subtract the tax-rate percentage from the total. At 20% UK VAT, a £600 gross is not £600 minus 20% = £480. The correct formula divides instead of subtracts: Pre-tax price = Total / (1 + Tax Rate). So at 20% VAT, the pre-tax base on a £600 gross is £600 / 1.20 = £500, with VAT of £100. For US sales tax at 8%, a $108 receipt total breaks down to $108 / 1.08 = $100 pre-tax with $8 of tax. Subtracting the tax rate from the total is one of the most common everyday math mistakes; it always gives an answer too low.

Tipping: A Globally Loaded Percentage

Tipping customs vary widely enough that an internationally read calculator should at least nod to the differences. United States. The post-pandemic working number sits at 18-20% of the pre-tax bill for table service, with 22-25% for exceptional service or higher-end venues. Tipping has a long and uncomfortable history: it spread in the United States in the late nineteenth century after the Civil War, when emancipated Black workers were hired into restaurant and hospitality jobs at no wage and forced to rely on patrons' gratuities for income. By the early 1900s, tipping was so widely seen as un-American that seven states passed laws to abolish it; by 1926 those laws had all been repealed because the practice had become impossible to police. There is now visible pushback: Square's processed-transaction data showed average restaurant tips falling from 15.5% in 2023 to 14.9% in Q2 2025, and a 2024 Bankrate survey found 63% of Americans now hold at least one negative view of tipping, up from 59% the year before. Europe. Service is essentially always already factored into European wages and often into menu prices. Rick Steves's frequently cited European travel guide gives the working norm: 5% is fully adequate at sit-down restaurants, 10% is generous, and "tipping 15 or 20 percent in Europe is unnecessary, if not culturally ignorant." Mediterranean menus often display servizio (Italian), service (French) or servicio (Spanish) lines explicitly. Japan. Tipping is not customary and can cause confusion or embarrassment when offered. The Japan National Tourism Organization is unambiguous: it is "not common to tip for services such as those provided in bars, cafes, restaurants, taxis, and hotels." The cultural backdrop is omotenashi, a hospitality ethos rooted in the Japanese tea ceremony, the entire point is to give service that needs no monetary postscript.

Probability and Percentage: The Base Rate Fallacy

Some of the most-cited failures of percentage reasoning happen in medicine, where doctors and patients try to translate test accuracy into the actually-relevant question: given a positive test, what is the probability the patient has the disease? The classic study is Eddy 1982. David Eddy posed American physicians a problem about mammography screening: 1% base rate of breast cancer in the screened population, 80% sensitivity (true positive rate), 9.6% false positive rate. A woman has just had a positive mammogram. What is the probability she has cancer? About 95 of 100 physicians answered around 75%. The correct answer, by Bayes's rule, is 7.7%. The doctors had confused the conditional probability "cancer given positive test" (the unknown, ≈ 7.7%) with the conditional probability "positive test given cancer" (the sensitivity, 80%). Gerd Gigerenzer's research has demonstrated repeatedly that the same problem becomes much easier when the percentages are restated as natural frequencies: concrete counts referring to the same denominator. Recast: 100 out of 10,000 women have cancer; 80 of those will get a positive mammogram; 950 of the remaining 9,900 (without cancer) will get a false positive. Now 80 + 950 = 1,030 women get a positive result, 80 actually have cancer, so the predictive value is 80/1,030 ≈ 7.8%. A meta-analysis shows accuracy of about 4% with conditional probabilities and about 24% with natural frequencies, a six-fold improvement. The Cochrane Collaboration now recommends natural-frequency framing for health-statistics communication. The lesson for percentages generally: whenever a question involves a conditional probability with a low base rate, mentally translating it into "X out of 1,000" tends to reduce error sharply.

Relative vs Absolute Risk Reduction

The relative-vs-absolute risk distinction is the same conceptual gap as percentage-points-vs-percent, recast for medicine. Imagine a drug trial where the placebo group has a 2% mortality rate and the treatment group has a 1% mortality rate. The absolute risk reduction is 2% − 1% = 1 percentage point. The relative risk reduction is 1% / 2% = 50%. The number needed to treat is 1 / 0.01 = 100, to prevent one death, you must treat 100 patients. The "50% reduction" headline is mathematically defensible but rhetorically lopsided: it suggests a much larger benefit than the underlying numbers support. A 2022 JAMA Internal Medicine meta-analysis of 21 randomised statin trials reported absolute risk reductions of 0.8% for all-cause mortality, 1.3% for myocardial infarction and 0.4% for stroke, and corresponding relative risk reductions of 9%, 29% and 14%. The relative numbers are the ones that get into headlines and pharmaceutical detailing materials; the absolute numbers are the ones that determine whether to take the drug. The asymmetry that critics most consistently flag is the practice of reporting benefits as relative risk reductions while reporting harms as absolute risk increases, a presentational choice that systematically inflates apparent benefit and downplays apparent harm. Relative changes are about ratios of probabilities and tell you how much a treatment moved the needle proportionally; absolute changes are about probabilities themselves and tell you how often the needle moved at all. They are both real, but only the absolute number translates directly into "one in N people will benefit."

Browser Arithmetic: Why 0.1 + 0.2 ≠ 0.3

A percentage calculator running entirely in the browser is at the mercy of JavaScript's number representation. Every JavaScript Number (other than BigInt) is stored as an IEEE 754 64-bit "double" with 53 bits of precision in the significand, about 15 to 17 significant decimal digits. The famous symptom is 0.1 + 0.2 === 0.3 returning false, because in binary, 0.1 is the infinite repeating fraction 0.0001100110011…; JavaScript truncates it to 53 bits, the same happens to 0.2, and adding them produces 0.30000000000000004. The same issue applies to most "round" decimal fractions: 0.7, 0.6, 0.3 all have non-terminating binary expansions. Number.MAX_SAFE_INTEGER equals 2⁵³ − 1 = 9,007,199,254,740,991, the largest integer JavaScript can represent exactly; beyond it, MAX_SAFE_INTEGER + 1 === MAX_SAFE_INTEGER + 2 evaluates to true, which is mathematically false. The most common trap a percentage calculator can fall into is toFixed: because numbers like 1.005 are not actually 1.005 in memory but 1.0049999…, the standard (1.005).toFixed(2) returns "1.00" rather than the expected "1.01". Library workarounds typically multiply by a power of ten, run Math.round, and divide back: Math.round(value * 100) / 100 is more predictable than raw toFixed. For money in particular, the safest approach is to do all arithmetic in integer cents (or pence, or øre), only converting back to currency-formatted strings at display time. The TC39 proposal-decimal effort is working toward a built-in Decimal type but is not yet shipping.

Banker's Rounding vs Round-Half-Up

The decimal-rounding question is also a policy question, not just a numerical one. Two main rounding rules compete in financial software. Round half up rounds 0.5 always up: 2.5 → 3, 3.5 → 4, 4.5 → 5. Round half to even (banker's rounding, Gaussian rounding) rounds 0.5 to the nearest even integer: 2.5 → 2, 3.5 → 4, 4.5 → 4. The motivation for round-half-to-even is to eliminate cumulative bias when many half-values are aggregated, round-half-up systematically rounds upward at the 0.5 boundary, so summing many rounded-up values produces a total higher than the true total. Round-half-to-even rounds up half the time and down half the time, leaving long-run bias near zero. It is the default rounding mode in IEEE 754 (what hardware floating-point operations do by default) and is widely used in financial systems for that reason. Cash rounding is a different beast: many countries round a cash transaction's total to the nearest available coin denomination because the smallest physical coin is no longer in circulation. Sweden pioneered the practice in 1972 (hence "Swedish rounding"); New Zealand followed in 1990; Canada eliminated the penny on 4 February 2013, with cash transactions there now rounding to the nearest C$0.05 while electronic payments stay precise to the cent. Australia, Finland, Ireland, Belgium, the Netherlands and Slovakia have all adopted variants.

When to Use Percentages, When to Use Raw Numbers

Edward Tufte's The Visual Display of Quantitative Information (1983, 2nd ed. 2001) lays out the canonical principles: maximise the data-ink ratio, watch the lie factor, and "above all else show the data." The percentage-versus-raw-number choice is a particular case of the lie-factor problem. A few principles fall out: when the base (denominator) is small or varies, percentages mislead, "a 100% increase in cases" is one case becoming two, alarming-sounding but trivial; always show the base. When the base is huge, raw numbers underwhelm, "1,000 deaths from a rare condition" sounds like a tragedy until you note the base is 100 million, making it 0.001%. When comparing across groups of different sizes, percentages are essential, but only if the underlying counts are large enough to make the percentage statistically meaningful; "60% of respondents" computed from a sample of five is noise dressed as signal. When the underlying quantity is itself a rate (interest, unemployment, vote share), use percentage points for changes and percent for ratios of changes. For very small or very large percentages, raw-frequency framing tends to communicate better than the percent. A percentage that is offered without its base, its sample size and its baseline comparison is a lie waiting to be told.

Frequently Asked Questions

What is the formula for percentage of a number?

To find X% of Y: multiply Y by X and divide by 100. The formula is (X / 100) × Y. So 15% of 200 = (15 / 100) × 200 = 30. The percent is the operator, the whole is the operand. Two of the trickier real-world variants: X% off a price means subtract (X/100) × Price from the price, so 20% off $50 leaves $40; X% tax on a price means add (X/100) × Price to the price, so 8% sales tax on $100 makes $108.

How do I calculate percentage change?

Percentage change = ((New − Old) / |Old|) × 100. If a stock goes from $50 to $65, that's ((65 − 50) / 50) × 100 = 30% increase. Note the asymmetry: the same dollar change gives different percentages depending on direction. Going from $65 back to $50 is a 23.1% decrease, not 30%, because the denominator changed. If a value goes up 25% then down 25%, you do not end at the start, you end at 0.9375 of where you started. Successive percentage changes multiply: (1 + r₁) × (1 + r₂) × …

What is the difference between percent and percentage points?

If an interest rate goes from 5% to 7%, that is a 2 percentage point increase but a 40% percent increase (because 2/5 = 0.40). Percentage points describe the absolute additive change in a rate; percent describes the relative multiplicative change. In financial markets the unit of account is the basis point (1 bp = 1/100th of a percentage point), so "the Fed raised rates by 25 basis points" means 0.25 percentage points and avoids the ambiguity entirely. Always use percentage points when reporting changes in things that are themselves rates (interest, unemployment, tax brackets, vote share).

How do I find the price before tax from a tax-inclusive total?

Divide, don't subtract. Pre-tax price = Total / (1 + Tax Rate). At 20% UK VAT, a £600 gross is £600 / 1.20 = £500 pre-tax with £100 of VAT. At 8% US sales tax, a $108 receipt is $108 / 1.08 = $100 pre-tax with $8 of tax. The intuitive subtraction "subtract 20% from £600 = £480" is wrong, that would leave you with the wrong base. The error always gives an answer too low.

What's the difference between markup and margin?

Markup is profit as a percentage of cost; margin is profit as a percentage of selling price. Sell something for $100 at a cost of $70: profit is $30, markup is $30/$70 = 42.9%, margin is $30/$100 = 30%. Markup is always larger than margin for the same item. Conversion: Margin = Markup / (1 + Markup) and Markup = Margin / (1 − Margin). So a 50% markup equals a 33.3% margin; a 50% margin equals a 100% markup. The classic small-business mistake is to set a price by adding what feels like a healthy markup and assume the resulting margin is the same number.

How precise are the calculations?

JavaScript stores all numbers as IEEE 754 doubles with about 15-17 significant decimal digits. For typical percentage work this is plenty. The known edge cases: very large integers above 2⁵³ − 1 lose precision; binary representations of 0.1, 0.2, 0.3 etc. are not exact, so 0.1 + 0.2 equals 0.30000000000000004 rather than 0.3. The displayed results are rounded sensibly (typically to 2-4 decimals depending on context). For exact financial arithmetic (payroll, tax computations, accounting) use a dedicated library that handles decimals natively (decimal.js, big.js) or work in integer cents/pence and only convert at display time.

Are my numbers sent anywhere?

No. All four calculation modes run entirely in your browser via JavaScript. The numbers you type never cross the network, verify in DevTools' Network tab while you compute, or take the page offline after it loads and confirm the calculator still works. Safe for figures involving salary, taxes, medical or financial details that you would not want copied onto a stranger's hard drive.

Related Tools