A continuous random variable is a variable that can take on any value within a range, including decimals and fractions, rather than only whole numbers or distinct categories. Think of measuring someone’s height: the result could be 170.1 cm, 170.15 cm, or 170.1527 cm. There’s no gap between possible values. This makes continuous random variables fundamentally different from discrete ones, which can only land on specific, countable outcomes like the number of heads in a coin toss.
What Makes a Variable “Continuous”
The key distinction is whether you’re measuring or counting. A continuous random variable takes values from an uncountably infinite set, meaning you could always squeeze another value between any two points. Height, weight, blood pressure, temperature, glucose levels, and age are all continuous variables. You measure them on a scale, and the precision is limited only by your instrument, not by the nature of the quantity itself.
A discrete random variable, by contrast, involves counting: the number of customers in a store, the result of rolling a die, the number of defective items in a shipment. You can list the possible outcomes. With a continuous variable, you can’t, because between any two values there are infinitely many others.
Why the Probability of Any Exact Value Is Zero
This is the most counterintuitive thing about continuous random variables: the probability that the variable equals any single, exact number is zero. Not approximately zero. Exactly zero.
The reason comes down to how probability works for continuous variables. Probability is represented as an area under a curve. When you ask for the probability of one specific point, you’re asking for the area of a line with no width, which is zero. It’s like asking for the area of a single line drawn on a sheet of paper. The line exists, but it has no measurable area.
This doesn’t mean the outcome is impossible. It means that with infinitely many possible values, no single value hogs any share of the probability. Instead, probability only makes sense over intervals. You can ask: what’s the probability that a person’s height falls between 165 cm and 175 cm? That question has a meaningful, nonzero answer.
The Probability Density Function
Since you can’t assign probabilities to individual values, continuous random variables use something called a probability density function (PDF). The PDF is a curve that describes how likely different ranges of outcomes are. Taller parts of the curve correspond to ranges where values are more concentrated; shorter parts mean values are less common there.
Two rules govern every valid PDF. First, the curve can never dip below zero, because negative probability doesn’t make sense. Second, the total area under the entire curve must equal exactly one. This reflects a simple fact: the variable has to take some value, so the combined probability across all possible outcomes is 100%.
To find the probability that your variable falls between two values, say between a and b, you calculate the area under the PDF curve between those two points. If you’ve taken calculus, this is the integral of the PDF from a to b. If you haven’t, just picture shading in the region under the curve between those boundaries and measuring that shaded area.
The Cumulative Distribution Function
The cumulative distribution function (CDF) answers a slightly different question: what’s the probability that the variable is less than or equal to some value x? It accumulates all the area under the PDF from the far left up to x. At the lowest possible value, the CDF starts at zero. At the highest, it reaches one.
The CDF gives you a convenient shortcut for interval probabilities. If you want the probability that a variable falls between a and b, you just subtract: take the CDF at b and subtract the CDF at a. This avoids recalculating the area from scratch each time. The PDF and CDF are two views of the same information. The CDF is the running total of the PDF, and the PDF is the rate of change of the CDF.
Mean and Spread
Just like discrete random variables, continuous ones have a mean (expected value) and variance that describe their center and spread. The expected value is a weighted average of all possible outcomes, where the weights come from the PDF. Conceptually, it’s the balance point of the curve: if you cut the shape out of cardboard, the expected value is where it would balance on your fingertip.
Variance measures how spread out the values are around that center. A small variance means most outcomes cluster tightly near the mean. A large variance means the distribution is wide and values routinely land far from the center. The square root of variance, called standard deviation, puts this spread back into the same units as the original variable, which makes it easier to interpret. If the mean height in a group is 170 cm with a standard deviation of 7 cm, most people fall within roughly 7 cm of the average.
Common Continuous Distributions
Several named distributions come up repeatedly because they describe patterns found throughout nature, engineering, and social science.
- Normal (Gaussian) distribution: The familiar bell curve. It shows up whenever many small, independent factors add together to produce an outcome. Heights, blood pressure readings, and measurement errors often follow this pattern. It’s symmetric, with most values near the mean and increasingly rare values in the tails.
- Uniform distribution: Every value in the range is equally likely. Imagine a spinner that’s just as likely to land at any angle. The PDF is a flat, horizontal line.
- Exponential distribution: Models the time between events that happen at a roughly constant rate, like the gap between customer arrivals or the lifespan of a lightbulb. It’s heavily skewed, with most values near zero and a long tail stretching to the right.
- Beta distribution: Useful for modeling proportions and probabilities themselves, since its values are confined between zero and one. It can take on a wide variety of shapes depending on its parameters.
- Gamma and chi-square distributions: These appear frequently in statistical testing and in modeling waiting times or total rainfall amounts. The chi-square distribution is a building block for many common statistical tests.
NIST catalogs over a dozen standard continuous distributions, each shaped by different real-world processes. Choosing the right one depends on what you’re modeling and what constraints the data naturally obeys.
Continuous vs. Discrete in Practice
The distinction matters because it changes how you analyze data. Continuous variables are compared using means, standard deviations, and tests designed for measured quantities (like the t-test). Discrete variables often require different tools, like counting how frequently each category appears.
Sometimes the boundary blurs. Age is technically continuous, but people often report it in whole years, making it behave like a discrete variable in datasets. Income is continuous in theory but gets grouped into brackets on surveys. In these cases, the choice of continuous vs. discrete treatment depends on how finely the data was actually recorded and what question you’re trying to answer.
What stays consistent is the core idea: a continuous random variable lives on a smooth number line with no gaps, its probabilities come from areas under a curve, and no single point on that line carries probability by itself. Once that clicks, the rest of probability and statistics for continuous data follows naturally from it.