How to Plot Error Bars in Matplotlib (plt.errorbar)

To plot error bars in Matplotlib, use plt.errorbar(x, y, yerr=errors); add xerr for horizontal error bars and capsize for the little caps. For a bar chart, pass yerr to plt.bar(). This guide covers plt.errorbar step by step: styling, error bars without a connecting line, asymmetric error bars, calculating the error (standard deviation or standard error) with pandas, bar charts, dates on the x-axis, and shaded error bands.

Tested with Python 3.12.5, Matplotlib 3.11.2, NumPy 2.5.3 and pandas 3.0.6. The charts are real Matplotlib windows and the table comes from the Windows Command Prompt. Reference: matplotlib.pyplot.errorbar.

Basic error bars with plt.errorbar()

import matplotlib.pyplot as plt

doses = [0, 10, 20, 30, 40, 50]
response = [2.1, 3.8, 5.9, 7.2, 8.1, 8.6]
error = [0.4, 0.5, 0.6, 0.5, 0.7, 0.6]

plt.errorbar(doses, response, yerr=error, capsize=4, marker="o")
plt.xlabel("Dose (mg)")
plt.ylabel("Response")
plt.title("plt.errorbar with vertical error bars")
plt.show()

Argument	What it does
yerr, xerr	Error size: one number for all points, one per point, or [lower, upper]
fmt	Marker and line format, e.g. "o" (markers only) or "-o"
capsize, capthick	Length and thickness of the caps
ecolor, elinewidth	Colour and width of the error bars
errorevery	Draw an error bar only on every n-th point
uplims, lolims	Show arrows for upper or lower limits

x and y error bars without a line

fmt="o" (or any marker without a line style) draws only the markers and the error bars, like a scatter plot with error bars:

import matplotlib.pyplot as plt
import numpy as np

rng = np.random.default_rng(4)
x = np.arange(1, 9)
y = 2 * x + rng.normal(0, 1.5, 8)
xerr = rng.uniform(0.2, 0.5, 8)
yerr = rng.uniform(0.8, 2.0, 8)

plt.errorbar(x, y, xerr=xerr, yerr=yerr,
             fmt="s",                      # markers only: no connecting line
             color="darkblue", ecolor="gray", elinewidth=1.5, capsize=5, capthick=1.5, markersize=6)
plt.title("x and y error bars, fmt='s' (no line)")
plt.grid(alpha=0.3)
plt.show()

Asymmetric error bars

When the error is not the same above and below the point, pass a list of two sequences: the distances below and the distances above:

import matplotlib.pyplot as plt

x = [1, 2, 3, 4, 5]
y = [10, 14, 12, 18, 16]
lower = [1, 2, 1.5, 3, 1]          # distance below each point
upper = [3, 1, 4, 2, 5]            # distance above each point

plt.errorbar(x, y, yerr=[lower, upper], fmt="o", capsize=5, color="tab:red")
plt.title("Asymmetric error bars: yerr=[lower, upper]")
plt.show()

Both lists contain positive distances, not the end values. If you have lower and upper bounds (for example a confidence interval), convert them first: lower = y - lo, upper = hi - y.

For more examples, see asymmetric error bars in Matplotlib.

Calculate the error: standard deviation or standard error

Error bars usually show the standard deviation (spread of the data) or the standard error of the mean (SEM, uncertainty of the mean). With pandas you can get both per group:

import numpy as np
import pandas as pd

rng = np.random.default_rng(10)
df = pd.DataFrame({
    "group": np.repeat(["Control", "Drug A", "Drug B"], 12),
    "score": np.concatenate([rng.normal(50, 6, 12), rng.normal(58, 8, 12), rng.normal(63, 5, 12)]),
})

summary = df.groupby("group")["score"].agg(["mean", "std", "count"])
summary["sem"] = summary["std"] / np.sqrt(summary["count"])      # standard error of the mean
print(summary.round(2))

Output:

          mean   std  count   sem
group                            
Control  48.85  4.11     12  1.19
Drug A   53.99  6.05     12  1.75
Drug B   62.89  5.22     12  1.51

Always say in the axis label or caption which one you plot: SEM bars are much shorter than standard-deviation bars for the same data.

Bar chart with error bars

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd

rng = np.random.default_rng(10)
df = pd.DataFrame({
    "group": np.repeat(["Control", "Drug A", "Drug B"], 12),
    "score": np.concatenate([rng.normal(50, 6, 12), rng.normal(58, 8, 12), rng.normal(63, 5, 12)]),
})
summary = df.groupby("group")["score"].agg(["mean", "std", "count"])
summary["sem"] = summary["std"] / np.sqrt(summary["count"])

fig, ax = plt.subplots(figsize=(6, 4.5))
ax.bar(summary.index, summary["mean"], yerr=summary["sem"], capsize=8,
       color=["gray", "tab:blue", "tab:green"], edgecolor="black")      # bar chart with error bars
ax.set_ylabel("Mean score (± SEM)")
plt.show()

Error bars with dates on the x-axis

errorbar() accepts datetime values for x, so time series with uncertainty work the same way:

import matplotlib.pyplot as plt
import matplotlib.dates as mdates
import pandas as pd

dates = pd.date_range("2025-01-01", periods=8, freq="W")
temperature = [3.1, 4.0, 2.5, 5.2, 6.8, 6.1, 8.4, 9.0]
spread = [1.2, 0.9, 1.5, 1.1, 1.4, 0.8, 1.0, 1.3]

fig, ax = plt.subplots(figsize=(8, 4))
ax.errorbar(dates, temperature, yerr=spread, fmt="-o", capsize=4)      # dates work as x values
ax.xaxis.set_major_formatter(mdates.DateFormatter("%d %b"))
ax.set_ylabel("Weekly mean temperature (°C)")
fig.autofmt_xdate()
plt.show()

Many points: errorevery and shaded error bands

With dense data, a bar at every point is hard to read. Draw one every n points with errorevery, or show the error as a shaded band with fill_between():

import matplotlib.pyplot as plt
import numpy as np

x = np.linspace(0, 10, 50)
y = np.sin(x) + x / 3
err = 0.2 + 0.05 * x

fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(10, 4))
ax1.errorbar(x, y, yerr=err, errorevery=5, capsize=3)                 # a bar on every 5th point
ax1.set_title("errorevery=5")
ax2.plot(x, y)
ax2.fill_between(x, y - err, y + err, alpha=0.3, label="± error")    # shaded error band
ax2.set_title("fill_between error band")
ax2.legend()
plt.tight_layout()
plt.show()

Error bars for several series

Shift each series slightly on the x-axis so the error bars do not hide each other:

import matplotlib.pyplot as plt
import numpy as np

weeks = np.arange(1, 7)
series = {"Model A": ([70, 74, 77, 81, 83, 86], [3, 3, 2, 2, 3, 2]),
          "Model B": ([68, 71, 76, 78, 84, 88], [4, 3, 3, 4, 3, 3])}

fig, ax = plt.subplots(figsize=(7, 4.5))
for shift, (name, (mean, err)) in zip([-0.1, 0.1], series.items()):
    ax.errorbar(weeks + shift, mean, yerr=err, fmt="o-", capsize=4, label=name)   # shift so bars do not overlap
ax.set_xlabel("Week")
ax.set_ylabel("Accuracy (%)")
ax.legend()
plt.show()

Related Matplotlib chart guides:

• Plot asymmetric error bars in Matplotlib
• Plot a bar chart in Matplotlib
• Plot multiple lines in Python
• Plot a line of best fit
• Set the x-axis label

Frequently asked questions