On this concept: Formula

Not sure which lesson to do first? Find your level and get a plan →

Math · Statistics & Probability · Grade 9-12

Normal Distribution

The normal distribution is a continuous, bell-shaped curve symmetric about its mean, where most values sit near the center and extremes are rare.

f(x)=1σ2π e−(x−μ)22σ2

Quick Answer

The normal distribution is a continuous, bell-shaped curve symmetric about its mean, where most values sit near the center and extremes are rare. Use it to model measurements that pile up around an average and thin out evenly on both sides, such as heights, measurement errors and many test scores. Two numbers describe it completely: the mean sets the center and the standard deviation sets the width. About 68% of values fall within 1 SD of the mean, 95% within 2 SDs and 99.7% within 3 SDs.

Why This Matters

Knowing data is roughly normal turns a mean and an SD into percentages. If IQ scores have mean 100 and SD 15, about 2.5% of people score above 130, and you can say so without seeing a single score. Apply the same rule to skewed data, like incomes, and the percentages come out badly wrong.

Intuitive Explanation

Drop thousands of balls through a Galton board, a board of pegs where each ball bounces left or right at random at every row. Most balls get a mix of lefts and rights and land near the middle; a ball that bounces the same way every time is rare and lands at an edge. The pile that forms is a bell. Heights behave similarly: a height is the sum of many small, separate influences, and sums of many small random pieces tend to be normal.

The bell's shape is fixed; only its center and width change. Change the mean and the bell slides along the axis; increase the SD and it spreads wider and flatter, because the total area under it is always 1. That fixed shape is why the 68-95-99.7 rule fits every normal distribution. If heights in a group have mean 162 cm and SD 7 cm, about 95% lie between 162−14=148 cm and 162+14=176 cm.

Not every single-peaked pile is normal. Household incomes peak at the low end and trail off in a long right tail, because a few very large incomes stretch it. For skewed data the mean sits away from the peak and 'about 95% within 2 SDs' can fail badly. Also, probability for a continuous variable is area: the height of the curve at 170 cm is not the chance of being exactly 170 cm, which is effectively 0.

Think of it like this. Picture where a steady archer's arrows land, left or right of the aim point. Most land close, misses to either side are equally common, and wild misses are rare: a normal curve. A shakier archer makes the same shape, only wider, like a bigger SD. Gusts that only ever blow from one side would push some arrows far that way and skew the pattern.

Core idea. The normal distribution is the bell-shaped curve where values cluster at the mean and thin out evenly on both sides.


When to Use

Use the normal model when a problem says the data are normally distributed or bell-shaped, gives a mean and standard deviation, and asks what percent or what probability falls in a range ('between 85 and 115', 'above 130'). Check the shape first: a roughly symmetric histogram with one peak supports it. For skewed or two-peaked data, describe it with the median and IQR. For cut-offs that are not whole SDs from the mean, convert to z-scores.

Tip. Ask: Is the data single-peaked, symmetric, and described by just a mean and a standard deviation?

How to Recognize It

Before using the normal model, look at how the values could be spread and what the question wants: the share, or the chance, of values in some range.

  1. Could the values pile up against a hard limit, such as waiting times that cannot go below 0?

    If yes, the data is likely right-skewed, with a long tail of large values, so normal-curve percentages will be off; use the median and IQR. If values can fall freely on either side of the mean, a normal model may fit.

  2. Does every value in the range look equally common, like the position of a clock's second hand when you glance at it?

    If yes, the data is uniform, a flat shape with no peak, and a probability is just the length of an interval divided by the whole range. A normal curve has a clear peak at the mean.

  3. Are the cut-offs a whole number of standard deviations from the mean, such as 1, 2 or 3?

    If yes, the 68-95-99.7 rule gives the share directly. If a cut-off falls in between, such as 1.4 standard deviations, convert it to a z-score and use the standard normal table or a calculator.

  4. Is the variable a count that takes only a few whole values, such as goals scored in a match?

    If yes, the normal curve usually fits poorly, since the values are few, whole and often lopsided. The normal model works best for measurements that can take any value, like times or lengths, or for large counts.

Normal Distribution compared with Skewed distribution, Uniform distribution, and Standard normal

Normal, skewed, uniform and standard normal are shapes a set of values can take, and the 68-95-99.7 rule fits only two of them. Ask whether the shape is a symmetric bell, and if so, whether it has been rescaled to mean 0 and standard deviation 1.

Normal Distribution

Meaning
Use this when data is single-peaked and symmetric about its mean and you want proportions or probabilities from a mean and SD.
Key test
Is the data single-peaked, symmetric, and described by just a mean and a standard deviation?
Formula
f(x)=1σ2π e−(x−μ)22σ2
Example
IQ scores are normal with mean 100 and SD 15. About what percent score between 85 and 115?

Skewed distribution

Meaning
An asymmetric pile with one long tail — not symmetric.
Key test
Use when data bunches on one side, like incomes or reaction times.
Example
Most people earn little, a few earn a lot

Uniform distribution

Meaning
Every outcome equally likely — flat, not bell-shaped.
Key test
Use when no value is more likely than another, like a fair die.
Example
Equal chance of each die face

Standard normal

Meaning
The special normal with mean 0 and SD 1, used after z-scoring.
Key test
Use after converting values to z-scores to read table probabilities.
Formula
Z∼N(0,1)
Example
Looking up P(Z<1.5)

Formula & Notation

f(x)=1σ2π e−(x−μ)22σ2
f(x)=1σ2πe−(x−μ)22σ2 for x∈(−∞,∞), with E(X)=μ and Var(X)=σ2

How to read it: X∼N(μ,σ2) reads 'X follows a normal distribution with mean μ and variance σ2'

Worked Examples

Example 1 — 68-95-99.7 estimate

Easy

IQ scores are normal with mean 100 and SD 15. About what percent score between 85 and 115?
Show the steps and the answer
  1. Measure the distances from the mean: 85=100−15 and 115=100+15.

    Both ends are exactly 1 SD from the mean, which matches the first number in the 68-95-99.7 rule.

  2. Confirm the rule applies: the problem states IQ scores are normal.

    The 68% figure is a property of the bell shape, so it needs normal data.

  3. Apply the rule: about 68% of values lie within 1 SD of the mean.

    For every normal distribution, the area between μ−σ and μ+σ is about 0.68.

  4. Split it by symmetry: about 34% score from 85 to 100 and about 34% from 100 to 115.

    Each side of the mean holds half of the 68%, which is useful when a question asks about only one side.

Answer: About 68%

Takeaway: When the bounds are whole SDs from the mean, the 68-95-99.7 rule gives the percent directly.

Example 2 — Right-skewed, not normal

Standard

Household incomes have a few very high earners and a long right tail — apply the 68-95-99.7 rule?
Show the steps and the answer
  1. Check the shape: a long right tail means the data is right-skewed, not symmetric.

    The 68-95-99.7 rule depends on the normal curve's symmetric shape.

  2. See what the tail does: if a town's incomes had mean $80k and SD $60k, 'mean ± 2 SD' would run from −$40k to $200k.

    No one earns a negative income, so the rule's interval cannot describe this data.

  3. Use the median and IQR instead.

    They depend on the order of the values, so a few very high incomes do not drag them.

Answer: Rule does not apply — use median/IQR

Takeaway: Check for symmetry before using the 68-95-99.7 rule; for skewed data, describe center and spread with the median and IQR.

Example 3 — Find the mistake: height is not probability

Application

Battery life is normally distributed with mean 50 hours and SD 4 hours, and the curve's peak height is about 0.10. Lee writes: "The peak height is 0.10, so 10% of batteries last exactly 50 hours." Find the mistake and correct it.
Show the steps and the answer
  1. Read the units of the height: 0.10 per hour.

    A curve's height is probability per unit of x; it becomes a probability only when multiplied by a width.

  2. Find the probability of exactly 50 hours: an interval from 50 to 50 has width 0, so its area is 0.

    For a continuous measurement, any single exact value has probability 0; only ranges of values have positive probability.

  3. See where Lee's 10% really belongs: the 1-hour window from 49.5 to 50.5 has area about 0.10×1=0.10.

    Height times width approximates area for a narrow strip, so about 10% of batteries last between 49.5 and 50.5 hours.

  4. Use an interval for a standard question: 46 to 54 hours is 50±4, so about 68% of batteries.

    That range is 1 SD on each side of the mean.

Answer: P(exactly 50)=0; about 10% last between 49.5 and 50.5 hours, and about 68% between 46 and 54 hours.

Takeaway: On a continuous curve, probability is the area over an interval; the height alone is a density, not a probability.

Where You'll See It

  • Scaled scores on standardized tests

    Many standardized tests scale scores to be roughly normal. If a section has mean 500 and SD 100, a score of 700 is 2 SDs above the mean, so only about 2.5% of test takers score higher. Score reports turn this into a percentile.

  • Designing clothing sizes

    If chest sizes in a population are roughly normal with mean 100 cm and SD 8 cm, a clothing maker knows about 95% of people fall between 84 and 116 cm. It can cover that range with a handful of standard sizes and leave the remaining 5% to special sizes.

Common Mistakes

  1. The slip: Applying the 68-95-99.7 rule to non-normal data

    Instead: the rule only holds for the symmetric bell.

  2. The slip: Assuming any single-peaked data is normal

    Instead: check for symmetry; a long tail means skew, not normal.

  3. The slip: Confusing the curve's height with probability

    Instead: for a continuous curve, probability is area under it over an interval, not the height at a point.


Mini Practice

Try these on your own. Tap Reveal when you want to check.

  1. Adult women's heights are normal with mean 165 cm and SD 7 cm. About what percent are between 151 and 179 cm?

    Hint: How many standard deviations from the mean is each end?

  2. Using the same heights (mean 165 cm, SD 7 cm), about what percent of women are taller than 172 cm?

    Hint: 172 is one SD above the mean. Split the leftover percent between the two tails.

  3. Battery life is normal with mean 40 hours and SD 4 hours. About what percent of batteries last between 36 and 48 hours?

    Hint: Split the interval at the mean and use half of each 68-95 piece.

  4. A machine fills cereal boxes with weights that are normal, mean 500 g, SD 5 g. Out of 2000 boxes, about how many weigh less than 490 g?

    Hint: Find the percent below 2 SDs under the mean, then apply it to 2000 boxes.

  5. A computer picks random numbers from 0 to 10, every value equally likely. Its histogram is flat. Is a normal model a good fit?

    Hint: Look for a single peak in the middle with tails that thin out.

  6. IQ scores are normal with mean 100 and SD 15. The curve's height at 100 is about 0.027, so Kim says "P(IQ=100)=2.7%." Correct her.

    Hint: For a continuous curve, probability comes from area over an interval.

Want the full set?

50 practice questions for this concept — free to try, every one with a complete worked solution showing the why, not just the answer.

Frequently Asked Questions

Why does the normal curve never quite reach the horizontal axis?

In the mathematical model, every value has some chance, however far it is from the mean, so the curve gets closer and closer to zero without arriving. The tails thin out fast: only about 0.3% of values lie beyond 3 standard deviations, and about 1 in 16,000 beyond 4. The model is an approximation, excellent near the middle and only roughly right far out in the tails.

Do real measurements follow the bell curve far out in the tails?

Often not. Daily stock market returns look roughly bell-shaped near the middle, but big crashes happen far more often than a normal model predicts. On 19 October 1987 the Dow Jones index fell about 22.6% in one day, a drop more than 20 standard deviations from the usual daily change, which a normal model says should essentially never happen. Data like this is called heavy-tailed, and normal percentages underestimate its extremes.

Why does the total of many dice make a bell shape?

One die gives a flat shape, since each face from 1 to 6 is equally likely. Two dice give a triangle that peaks at 7, because more pairs make 7 than any other total. With 10 dice the totals already form a close copy of a bell centered at 35. Adding many independent random pieces tends to produce a normal shape whatever the pieces look like, a fact called the central limit theorem, which is why the bell curve turns up so often.

Learning Path

Normal Distribution

You are here

You need the Mean, which sets the center of the bell, and Standard Deviation, which sets its width and is the unit of the 68-95-99.7 rule. The normal distribution adds a model that turns those two numbers into percentages. Next, the Z-Score measures any value's distance from the mean in SDs so one standard table works for every bell, and the Central Limit Theorem explains why sample averages are normal even when the data are not.

See Also