A normal distribution calculator answers “what fraction of values lie below x?”. The inverse normal calculator answers the opposite question: “below which value do a given fraction of values lie?”. It is the tool for setting cut-offs: the score needed to reach the top 10%, the weight below which only 2.5% of packages fall, or the interval that contains the central 95% of a population.
Mathematically it uses the inverse of the standard normal cumulative distribution, Φ⁻¹, also called the probit or quantile function. Φ⁻¹(p) is the z-score with area p to its left; multiplying by σ and adding μ moves it onto your scale.
Choose the kind of area you have: left tail P(X < x) = p, right tail P(X > x) = p, both tails together P(|X − μ| > d) = p, or the central area P(|X − μ| < d) = p, which is how confidence levels are written. For the two symmetric options the result is a pair of values x₁ = μ − d and x₂ = μ + d.
How to use the Inverse Normal Distribution Calculator
- Probability (area)
- The probability is the area
- Mean (μ)
- Standard deviation (σ)
Everything is calculated in your browser, so results appear instantly and nothing you type is sent to a server.
How it works
The calculator inverts the standard normal CDF numerically to get the z-score for the requested area, then converts it with x = μ + σz. For two-sided areas it splits the probability equally between the tails.
Worked example
A test is normally distributed with mean 100 and standard deviation 15. Which scores mark the central 80% of results? Choose “inside μ ± d” and p = 80%. Then d = 15 × Φ⁻¹(0.90) = 15 × 1.2816 = 19.22, so the central 80% lies between 80.78 and 119.22. The same pair answers “outside μ ± d” with p = 20%.
Common critical values
| Area | One tail z | Two tails / central z |
|---|---|---|
| 90% central (10% outside) | 1.2816 | 1.6449 |
| 95% central (5% outside) | 1.6449 | 1.9600 |
| 98% central (2% outside) | 2.0537 | 2.3263 |
| 99% central (1% outside) | 2.3263 | 2.5758 |
| 99.9% central | 3.0902 | 3.2905 |
Formula
x = μ + σ · Φ⁻¹(p) (left tail) · x = μ + σ · Φ⁻¹(1 − p) (right tail) · d = σ · Φ⁻¹(1 − p/2) (two tails) · d = σ · Φ⁻¹((1 + p)/2) (central area)
Worked example
Inputs: Probability (area): 90 %; The probability is the area: Left of x – P(X < x) = p; Mean (μ): 100; Standard deviation (σ): 15.
Results: Result = x = 119.2233 – P(X < x) = 90%; Value x (lower value x₁ for two-sided areas) = 119.2233; z-score of x (x₁) = 1.2816; Percentile of x (x₁) = 90 %.
Key terms
- Quantile function
- The inverse of the CDF: the value with a given area to its left.
- Φ⁻¹ (probit)
- The inverse of the standard normal CDF, giving the z-score for an area.
- Tail area
- The probability in one or both ends of the distribution beyond a cut-off.
- Critical value
- A z-score used as a cut-off in tests and confidence intervals, such as 1.96.
Frequently asked questions
What z-score is the 95th percentile?
Φ⁻¹(0.95) = 1.645. For the 97.5th percentile it is 1.960 and for the 99th 2.326.
What IQ is needed for the top 10%?
With μ = 100 and σ = 15: x = 100 + 15 × 1.2816 = 119.22.
What score marks the bottom 20%?
x = 100 + 15 × Φ⁻¹(0.20) = 100 − 15 × 0.8416 = 87.38.
How do I find a 95% interval around the mean?
Choose "inside μ ± d" with p = 95%: d = 1.96σ, so for μ = 100, σ = 15 the interval is 70.60 to 129.40.
What value cuts off the lowest 30% of N(165, 34²)?
z = Φ⁻¹(0.30) = −0.5244, so x = 165 − 34 × 0.5244 = 147.17.
Can I use it for a lognormal variable?
Yes. Find the quantile of ln X with μ and σ of the logs, then take eˣ of the answer.
Why can't p be exactly 0% or 100%?
The normal curve never reaches zero height, so those areas correspond to minus or plus infinity.