normalpdf vs normalcdf: which to use

By Jude Wallis · Published

normalcdf(lower, upper, mean, SD) returns the area under the normal curve between two values, and that area is a probability. normalpdf(x, mean, SD) returns the height of the curve at a single value, which is not a probability. For probability questions, use normalcdf.

AP Statistics: Unit 2 (topics 2.11 The Normal Distribution). Finding probabilities as areas under a normal curve is Unit 2 topic 2.11, The Normal Distribution, in the Fall 2026 AP Statistics course. normalpdf and normalcdf are TI-84 command names for two calculations on that curve. The topic concerns areas, which is the normalcdf calculation.

normalpdf vs normalcdf: the short answer

Both commands are in the TI-84 distribution menu (2nd VARS), and they return different kinds of numbers.

  • normalcdf returns an area. You give it a lower bound and an upper bound, and it returns the probability that a normal variable falls between them.
  • normalpdf returns a height. You give it one value of xx, and it returns how high the normal curve is at that point.

For a normal variable, probability is area under the curve. A height alone has no width, so it is not an area and not a probability. Every question that asks for a probability, a proportion, or a percent under a normal model is a normalcdf question.

What normalcdf does

The syntax is normalcdf(lower bound, upper bound, mean, standard deviation). It returns P(lower<X<upper)P(\text{lower} < X < \text{upper}) for a normal variable XX with that mean μ\mu and standard deviation σ\sigma. The answer is always between 0 and 1.

Suppose adult women's heights in one population are normal with μ=64\mu = 64 inches and σ=2.5\sigma = 2.5 inches. The probability a randomly chosen woman is between 61.5 and 66.5 inches is normalcdf(61.5, 66.5, 64, 2.5) =0.6827= 0.6827. Those bounds are one standard deviation each side of the mean, so the result is the 68% of the empirical rule.

For a one-sided question, put a very large number on the open side: normalcdf(69, 1E99, 64, 2.5) =0.0228= 0.0228 is the probability of a height above 69 inches. If you leave out the mean and standard deviation, the calculator uses the standard normal, μ=0\mu = 0 and σ=1\sigma = 1, and reads the bounds as z-scores. Full keystrokes are in normalcdf and invNorm on the TI-84.

What normalpdf does

The syntax is normalpdf(x, mean, standard deviation). It evaluates the normal density function at xx:

f(x)=1σ2π e−12(x−μσ)2f(x) = \frac{1}{\sigma\sqrt{2\pi}} \, e^{-\frac{1}{2}\left(\frac{x - \mu}{\sigma}\right)^2}

The result is the height of the bell curve above that xx. For the heights above, normalpdf(64, 64, 2.5) =0.1596= 0.1596 is the height of the curve at its peak. It does not mean that 15.96% of women are 64 inches tall. For a continuous variable, the probability of any one exact value is 0.

A density height can even be greater than 1, which no probability can be. A normal curve with σ=0.1\sigma = 0.1 has a peak height of normalpdf(0, 0, 0.1) =3.9894= 3.9894. The curve is tall because it is narrow, and the area under it is still 1.

The main use of normalpdf is drawing the curve. Entering Y1 = normalpdf(X, 64, 2.5) in the Y= editor graphs the bell shape. It rarely appears in a calculation. The idea behind the height is explained in what is a density curve.

The differences side by side

Featurenormalpdfnormalcdf
Syntaxnormalpdf(x, μ\mu, σ\sigma)normalcdf(lower, upper, μ\mu, σ\sigma)
Number of x-values enteredOneTwo
ReturnsHeight of the curve at xxArea under the curve between the bounds
Is the output a probabilityNoYes
Range of the outputAny positive number, can exceed 1Between 0 and 1
Typical useGraphing the normal curveFinding probabilities, proportions, and percents
With μ=64\mu = 64, σ=2.5\sigma = 2.5normalpdf(64, 64, 2.5) =0.1596= 0.1596normalcdf(61.5, 66.5, 64, 2.5) =0.6827= 0.6827

The letters carry the meaning. In pdf, the d stands for density, a height. In cdf, the c stands for cumulative, an accumulated area.

Why the mix-up happens: binompdf is different

The confusion usually starts with the binomial commands. binompdf(n, p, x) does return a probability: binompdf(10, 0.5, 5) =0.2461= 0.2461 is the probability of exactly 5 successes in 10 trials. A binomial variable is discrete, so each whole-number count has its own bar, and the height of the bar is the probability of that count.

A normal variable is continuous. There are no bars, only a smooth curve, and probability comes from area over an interval. So the same three letters mean two different things.

CommandType of variableIs the output a probability
binompdfDiscreteYes, P(X=x)P(X = x)
binomcdfDiscreteYes, P(X≤x)P(X \le x)
normalpdfContinuousNo, a curve height
normalcdfContinuousYes, P(lower<X<upper)P(\text{lower} < X < \text{upper})

The binomial pair is covered in binompdf vs binomcdf.

The classic mix-up and how to avoid it

The error is typing normalpdf when the question asks for a probability. The trouble is that the wrong answer often looks reasonable. For the heights above, the probability of being taller than 69 inches is normalcdf(69, 1E99, 64, 2.5) =0.0228= 0.0228. Typing normalpdf(69, 64, 2.5) returns 0.02160.0216, a believable-looking number that is the curve's height at 69 and answers no probability question.

Three checks catch it.

  1. Count the boundaries. A probability under a normal curve needs two bounds, even if one is 1E99 or -1E99. If you entered a single xx-value, you used the wrong command.
  2. Sketch and shade. Draw the curve and shade the region the question describes. A shaded region is an area, which means normalcdf.
  3. Check "exactly" questions. If a normal problem asks for the probability of exactly one value, the answer is 0. No command is needed.

When written work is graded, a calculator command is not a full answer. Name the distribution with its mean and standard deviation, write the probability statement such as P(X>69)P(X > 69), then give the value.

One distribution, both commands

Heights of adult women in a population are normal with mean 64 inches and standard deviation 2.5 inches. Find the probability that a randomly chosen woman is taller than 69 inches. Then find what normalpdf returns at 69 and explain what that number is.

  1. Write the probability statement: P(X>69)P(X > 69) where XX is normal with μ=64\mu = 64 and σ=2.5\sigma = 2.5.

  2. This is an area to the right of 69, so use normalcdf with a very large upper bound: normalcdf(69, 1E99, 64, 2.5) =0.0228= 0.0228.

  3. Check with a z-score: z=69−642.5=2z = \frac{69 - 64}{2.5} = 2. The area to the left of z=2z = 2 in the z-table is 0.97720.9772, so the area to the right is 1−0.9772=0.02281 - 0.9772 = 0.0228. The two methods agree.

  4. Now evaluate the density: normalpdf(69, 64, 2.5) =0.0216= 0.0216.

  5. Interpret that value. It is the height of the normal curve at x=69x = 69. It is close to 0.02280.0228 by coincidence and has a different meaning.

  6. Connect height to area. Height times a small width approximates the area of a thin strip. The strip from 68.5 to 69.5 has width 1, so its area is about 0.0216×10.0216 \times 1. The calculator gives normalcdf(68.5, 69.5, 64, 2.5) =0.0220= 0.0220, which is close.

P(X>69)=0.0228P(X > 69) = 0.0228, from normalcdf(69, 1E99, 64, 2.5). The value normalpdf(69, 64, 2.5) =0.0216= 0.0216 is the height of the density curve at 69 inches. It is not the probability of being taller than 69 inches, and it is not the probability of being exactly 69 inches, which is 0.

A density height greater than 1

A machine fills bottles with a fill error that is normal with mean 0 milliliters and standard deviation 0.1 milliliters. Find normalpdf at 0, and find the probability that the error is between -0.1 and 0.1 milliliters. Explain why the first number cannot be a probability.

  1. Evaluate the density at the mean: normalpdf(0, 0, 0.1) =10.12π≈3.9894= \frac{1}{0.1\sqrt{2\pi}} \approx 3.9894.

  2. Compare with the standard normal curve, whose peak height is normalpdf(0, 0, 1) ≈0.3989\approx 0.3989. Dividing σ\sigma by 10 makes the curve 10 times narrower, so its peak is 10 times taller: 0.3989×10=3.9890.3989 \times 10 = 3.989.

  3. Find the probability for the interval: normalcdf(-0.1, 0.1, 0, 0.1) =0.6827= 0.6827. The bounds are one standard deviation each side of the mean.

  4. Compare the two outputs. The height 3.98943.9894 is greater than 1, and no probability is greater than 1. The area 0.68270.6827 is between 0 and 1, as every probability is.

normalpdf(0, 0, 0.1) ≈3.9894\approx 3.9894 is a curve height, and since it exceeds 1 it cannot be a probability. The probability that the error is within 0.1 milliliters of zero is normalcdf(-0.1, 0.1, 0, 0.1) =0.6827= 0.6827.

Frequently asked questions

When would I ever use normalpdf?

Mostly to graph a normal curve, by entering Y1 = normalpdf(X, mean, SD) in the Y= editor. It gives the curve's height at each x-value. For probabilities, proportions, and percents, use normalcdf.

Why does normalpdf not give a probability when binompdf does?

A binomial variable is discrete, so each count has a bar whose height is its probability. A normal variable is continuous, so probability is area over an interval, and the height at one point is a density. A single exact value of a continuous variable has probability 0.

What do I enter for the lower bound if the question says "less than"?

Use a very large negative number such as -1E99, entered with the EE key. For example, normalcdf(-1E99, 61.5, 64, 2.5) returns 0.1587, the probability of a value below 61.5 when the mean is 64 and the standard deviation is 2.5.