Home › Math Lessons › Statistics › Normal Distribution Explained
StatisticsBeginner

Normal Distribution Explained

The bell curve shows up everywhere , heights, test scores, measurement errors, blood pressure readings. Understanding why so many things follow this shape, and what you can do with that knowledge, is one of the most useful things statistics can teach you.

✔ The short version

A normal distribution is a symmetric, bell-shaped curve where most values cluster near the middle and fewer values appear as you move toward the extremes. It's defined by just two numbers: the mean (where the center sits) and the standard deviation (how wide the bell is). The 68-95-99.7 rule tells you what percentage of data falls within 1, 2, and 3 standard deviations.

What the bell curve actually shows

Picture a histogram of adult male heights. Most men are somewhere in the middle , around 5'9" or 5'10". Fewer men are 5'5" or 6'1". Very few are 5'0" or 6'6". The histogram forms a hill shape: tall in the middle, tapering symmetrically on both sides. That's a normal distribution.

The shape is symmetric around the mean. That means exactly half the values fall above the mean and half fall below. The mean, median, and mode are all the same number, sitting right at the peak of the bell.

Width varies. A tight bell curve with a small standard deviation means values are packed closely around the mean. A wide, flat curve with a large standard deviation means values are spread out. Same shape, just stretched or compressed horizontally.

The rule that makes the normal distribution practical

For any normal distribution, regardless of what the mean and SD actually are, the same proportions of data fall within 1, 2, and 3 standard deviations of the mean. This is the 68-95-99.7 rule, sometimes called the empirical rule.

68% of values fall within 1 SD of the mean. 95% within 2 SD. 99.7% within 3 SD. That leaves only 0.3% outside three standard deviations , the extreme tails.

IQ scores are calibrated to mean 100, SD 15. So 68% of people score between 85 and 115. 95% between 70 and 130. Getting a score above 145 puts you in the top 0.15% of the population. This isn't a claim about intelligence , it's just how the test was designed. But it shows the rule in action.

Applying the rule to a real dataset
A manufacturer produces bolts with target diameter 10mm and standard deviation 0.2mm.
168% of bolts measure between 9.8mm and 10.2mm (within 1 SD)
295% measure between 9.6mm and 10.4mm (within 2 SD)
3Only 0.3% fall outside 9.4mm to 10.6mm (beyond 3 SD)
If tolerance is ±0.4mm, about 95% of bolts pass. If tolerance is ±0.2mm, only about 68% pass.

Why does this shape appear so often?

The Central Limit Theorem, one of the most important results in probability theory, answers this. It says that when you add together many independent random variables, the result tends toward a normal distribution , regardless of what distribution those individual variables follow. This is a remarkable mathematical fact.

Human height depends on hundreds of genetic and environmental factors. Each contributes a small random amount. When you add many small independent random effects together, you get a normal distribution. Same logic applies to measurement errors (sum of many small disturbances), exam scores (sum of many small knowledge factors), and countless other things.

The Central Limit Theorem is also why sample means are normally distributed even when the underlying data isn't. If you repeatedly sample from any population and take the mean each time, those means will form a normal distribution. This property is what makes most of classical statistics work.

When the normal distribution does not apply

Not everything is normally distributed, and assuming it is when it isn't causes problems.

Income is not normally distributed , it's right-skewed, with a long tail of very high earners. Using the normal distribution to model income gives wildly wrong predictions about the rich end of the distribution. This matters for insurance, tax policy, and economics.

Financial returns are not perfectly normal either. The tails are "fatter" than a normal distribution predicts , extreme events happen more often than normal distribution models suggest. The 2008 financial crisis partly happened because models assumed normality and wildly underestimated the probability of extreme losses.

Before applying normal distribution assumptions, it's worth checking whether your data is actually approximately bell-shaped. A histogram is usually enough to tell.

Practice Problems

A dataset is normally distributed with mean 50 and SD 10. What percentage of values fall between 30 and 70?
30 and 70 are each 2 standard deviations from the mean (50±20). By the 68-95-99.7 rule, 95% of values fall within 2 SDs of the mean.
A student scores 1 SD below the mean on an exam. What percentile are they approximately in?
One SD below the mean puts you at the 16th percentile approximately. Here's why: 68% of students score within 1 SD of the mean, leaving 32% outside. Half of those (16%) are below 1 SD. So the student is at roughly the 16th percentile.
Why can't a perfectly normal distribution describe something like income?
Normal distributions are symmetric and extend to negative infinity. Income can't be negative (it has a hard floor at zero) and is heavily right-skewed , a few very high earners pull the mean well above the median. Income follows a log-normal distribution instead.