How to Create a Bell Curve in Excel (Step by Step)
Build a bell curve in Excel the modern way with NORM.DIST and dynamic arrays — one spill formula, no fill handle. Free working .xlsx included.
How to Create a Bell Curve in Excel (Step by Step)
A bell curve answers a question a bar chart can’t: not just what your numbers are, but how they’re distributed. Whether you’re reviewing test scores, employee performance ratings, product weights, or daily sales, the curve shows you where the middle sits, how spread out results are, and whether that one outlier really is an outlier.
Most Excel tutorials still teach this the 2010 way — typing out a series of x-values by hand and dragging a fill handle down 60 rows. Microsoft’s own support article goes further back and recommends the Analysis ToolPak. You don’t need any of that. In current Excel, two formulas build the entire curve, and it updates itself when your data changes.
This guide covers the modern dynamic-array method first, then a clearly labeled legacy method if you’re on Excel 2019 or earlier. There’s a free working file at the end so you can pull the finished chart apart. (Not sure a bell curve is the right chart for your data in the first place? Start with our chart decision guide.)
What You Need Before You Chart Anything
A bell curve is a plot of the normal distribution, and a normal distribution is defined by exactly two numbers:
- The mean — the average of your data. Calculate it with
=AVERAGE(A2:A41). - The standard deviation — how spread out your data is. Calculate it with
=STDEV.S(A2:A41)for a sample, or=STDEV.P(A2:A41)if your data is the entire population.
Which standard deviation function? If your 40 scores are all the scores that exist — every student in the class — use STDEV.P. If they’re a sample of something bigger — 40 customers out of thousands — use STDEV.S. For most business data, STDEV.S is the right call. The difference is small with large datasets, but it’s the kind of detail that gets your analysis questioned in a review, so get it right.
Put your raw data in column A, your mean in E2, and your standard deviation in E3. Label them. Future-you will thank present-you.
The Modern Method: Two Formulas, Zero Dragging
This works in Excel for Microsoft 365 and Excel 2021 or later — any version with dynamic arrays.
Step 1: Generate the x-values with SEQUENCE
The curve needs a range of x-values running from about three standard deviations below the mean to three above. That range covers 99.7% of a normal distribution, which is all anyone needs to see.
In cell G2, enter:
| E | F | G | |
|---|---|---|---|
| 1 | Mean | x-values | |
| 2 | 72.5 | 45.2 | |
| 3 | SD | 46.1 | |
| 4 | 9.1 | 47.0 | |
| 5 | 47.9 | ||
| 6 | 48.8 |
=SEQUENCE(61, 1, E2-3*E3, E3/10)Read it left to right: 61 values, in one column, starting at mean minus three standard deviations, stepping up by one-tenth of a standard deviation each time. Press Enter and the whole series spills down automatically — no fill handle, no typing 35, 36, 37 down the page.
Step 2: Calculate the curve with NORM.DIST
In cell H2, enter:
=NORM.DIST(G2#, E2, E3, FALSE)The G2# reference (that’s G2 followed by a hash) tells Excel to use the entire spilled range from Step 1, so this single formula calculates every point on the curve at once.
The FALSE at the end matters more than any other character in this formula. FALSE gives you the probability density function — the bell shape. TRUE gives you the cumulative distribution — an S-curve that climbs from 0 to 1. If your “bell curve” comes out looking like a ski slope, this argument is why.
Step 3: Insert a scatter chart with smooth lines
- Select both spilled ranges (G2:H62 — or click G2, then Ctrl+Shift+End if they’re the only data there).
- Go to Insert > Charts > Insert Scatter (X, Y) or Bubble Chart > Scatter with Smooth Lines.
- Excel draws the bell.
Don’t use a line chart here. A line chart treats your x-values as category labels spaced evenly, which happens to look fine with evenly stepped values but breaks the moment you change the step. A scatter chart plots x-values as actual numbers, which is what a distribution is.
Step 4: Format it like you mean it
The default chart is serviceable. Three minutes makes it presentable:
- Delete the gridlines and legend. One series doesn’t need a legend, and gridlines add nothing to a smooth curve.
- Fix the x-axis bounds. Right-click the x-axis > Format Axis, and set the minimum and maximum to match your SEQUENCE start and end. This stops Excel padding the chart with empty space.
- Title it with the finding, not the file name. “Q2 Assessment Scores: Distribution” beats “Chart 1.”
The payoff of the dynamic method shows up now: change any score in column A and the mean, standard deviation, x-values, curve values, and chart all update instantly. If your chart isn’t updating when the data changes, that’s a separate (and fixable) problem — see Excel chart not updating? 7 fixes.
Legacy Method: For Excel 2019 and Earlier
This is the fallback, not the recommendation. If your Excel doesn’t support SEQUENCE, here’s the manual version:
- Calculate your mean and standard deviation as above.
- In G2, type your starting x-value:
=E2-3*E3. Copy the result as a value if you prefer, or in G3 enter=G2+$E$3/10and drag the fill handle down to row 62. - In H2, enter
=NORM.DIST(G2, $E$2, $E$3, FALSE)— note the dollar signs locking the mean and standard deviation — and drag it down alongside. - Select both columns and insert a scatter with smooth lines, exactly as in the modern method.
It produces an identical chart. It just won’t resize itself gracefully, and every edit means re-dragging formulas. If you’re doing this more than twice a year, that’s an argument for upgrading Excel.
Overlaying Your Actual Data (the Step Worth Adding)
A theoretical curve on its own is only half the story. The interesting question is usually how well your real data fits it. The quick version:
- Build a small frequency table: bins in one column (say, every 5 points), and
=FREQUENCY(A2:A41, bins)spilled next to it. - Add the frequencies to your chart as a second series, change its chart type to a clustered column, and put it on the secondary axis.
Now you can see at a glance whether your scores are genuinely bell-shaped or skewed — which is frequently the more useful finding. A “bell curve” chart of data that isn’t normally distributed is a confident-looking lie, and it’s a mistake worth actively checking for rather than assuming away.
Common Mistakes That Ruin Bell Curves
- Using TRUE instead of FALSE in NORM.DIST. You get an S-curve. Covered above, repeated here because it’s the #1 support question on this topic.
- Using a line chart instead of a scatter chart. Works by accident until it doesn’t.
- STDEV.P on sample data. Slightly understates the spread; makes your curve look tighter than reality.
- Charting the curve of data that isn’t normal. Salaries, house prices, and response times are usually right-skewed. Plot the actual frequency distribution first; fit the curve second.
FAQ
Can Excel make a bell curve automatically? Not in one click, but the modern method above is two formulas and one chart insert — under two minutes. If you’d rather describe the chart in plain English and have AI draft the formulas, our ChatGPT prompts for Excel charts includes prompts for exactly this build.
What’s the difference between NORM.DIST and NORMDIST?
NORMDIST (no dot) is the pre-2010 version, kept for compatibility. Use NORM.DIST. They calculate identically.
How many x-value points do I need? The 61 points from our SEQUENCE formula produce a perfectly smooth curve. Fewer than about 30 and you’ll see the curve get angular; more than 100 is wasted effort.
Can I shade the area under part of the curve? Yes — add a second NORM.DIST series covering only the x-range you want to highlight, change its chart type to area, and set a transparent fill. It’s the standard way to visualize “the middle 68%” or a pass threshold.
Download the Working File
The free workbook below contains the finished dynamic-array bell curve, the sample data, and the legacy version on a separate sheet — every formula live, so you can swap in your own numbers and watch the curve rebuild itself.