How to Make a Box Plot in Python (Matplotlib & Seaborn Guide)

A box plot (also called a box-and-whisker plot) is the fastest way to compare the distribution of a numeric variable across groups. It shows the median, the interquartile range (IQR), and outliers in a single compact glyph, which makes it a staple of biology, clinical, and engineering papers. This guide covers everything from a one-line box plot to a fully styled, journal-ready figure in both matplotlib and seaborn.

If you are still deciding whether a box plot is the right chart at all, read our box plot vs violin plot vs bar chart comparison first. Otherwise, let us build one.

1. The minimal box plot

Pass a list of arrays to ax.boxplot(). Each array becomes one box. Use tick_labels to label the groups. On matplotlib 3.9+ the argument is tick_labels; on older versions it was labels.

import matplotlib.pyplot as plt
import numpy as np

rng = np.random.default_rng(42)
control = rng.normal(100, 8, 60)
treatment = rng.normal(88, 11, 60)

fig, ax = plt.subplots(figsize=(6, 4.5))
ax.boxplot([control, treatment], tick_labels=["Control", "Treatment"])
ax.set_ylabel("Cell viability (%)")
ax.set_title("Basic box plot")
plt.tight_layout()

2. Color and style the boxes

Filled boxes require patch_artist=True. The boxplot() call returns a dictionary of artists, so you can iterate over bp["boxes"] to set a per-group color. Keep the median line dark for contrast and reduce flier marker size so outliers do not dominate.

import matplotlib.pyplot as plt
import numpy as np

rng = np.random.default_rng(42)
groups = [rng.normal(loc, 9, 60) for loc in (100, 88, 72)]
labels = ["Control", "Drug A", "Drug B"]
colors = ["#6b7280", "#7c3aed", "#ec4899"]

fig, ax = plt.subplots(figsize=(6.2, 4.6))
bp = ax.boxplot(
    groups,
    tick_labels=labels,
    patch_artist=True,       # required to fill the boxes
    widths=0.55,
    medianprops=dict(color="black", linewidth=1.4),
    flierprops=dict(marker="o", markersize=4, markerfacecolor="none"),
)

for patch, color in zip(bp["boxes"], colors):
    patch.set_facecolor(color)
    patch.set_alpha(0.65)

ax.set_ylabel("Cell viability (%)")
ax.spines[["top", "right"]].set_visible(False)
plt.tight_layout()

3. The faster route: seaborn

Seaborn works directly from a tidy (long-format) DataFrame and handles grouping, coloring, and labels for you. Pass the grouping column to both x and hue (with legend=False) to get per-group colors without a redundant legend, which is the current seaborn-recommended pattern.

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

rng = np.random.default_rng(42)
df = pd.DataFrame({
    "viability": np.concatenate([rng.normal(loc, 9, 60) for loc in (100, 88, 72)]),
    "group": np.repeat(["Control", "Drug A", "Drug B"], 60),
})

fig, ax = plt.subplots(figsize=(6.2, 4.6))
sns.boxplot(data=df, x="group", y="viability", hue="group",
            palette=["#6b7280", "#7c3aed", "#ec4899"], legend=False, ax=ax)
ax.set(xlabel="", ylabel="Cell viability (%)")
sns.despine()
plt.tight_layout()

4. Overlay the raw data points

Most journals now ask you to show individual observations alongside the summary, especially for small samples. Combine sns.boxplot (with fliersize=0 so outliers are not drawn twice) and sns.stripplot. This is the single most reviewer-friendly upgrade you can make to a box plot.

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

rng = np.random.default_rng(7)
df = pd.DataFrame({
    "value": np.concatenate([rng.normal(loc, 9, 18) for loc in (100, 88, 72)]),
    "group": np.repeat(["Control", "Drug A", "Drug B"], 18),
})

fig, ax = plt.subplots(figsize=(6.2, 4.6))
sns.boxplot(data=df, x="group", y="value", hue="group",
            palette="Greys", legend=False, fliersize=0, ax=ax)
sns.stripplot(data=df, x="group", y="value", color="#7c3aed",
              size=4, jitter=0.18, alpha=0.85, ax=ax)
ax.set(xlabel="", ylabel="Response")
sns.despine()
plt.tight_layout()

5. Horizontal box plots for long labels

When group names are long (site names, gene IDs, drug names), a horizontal layout keeps them readable. In matplotlib, set vert=False; in seaborn, swap the x and y arguments.

import matplotlib.pyplot as plt
import numpy as np

rng = np.random.default_rng(42)
labels = ["Site E", "Site D", "Site C", "Site B", "Site A"]
data = [rng.normal(loc, 6, 50) for loc in (40, 46, 52, 58, 63)]

fig, ax = plt.subplots(figsize=(6.5, 4.2))
ax.boxplot(data, tick_labels=labels, vert=False, patch_artist=True,
           boxprops=dict(facecolor="#7c3aed", alpha=0.6))
ax.set_xlabel("Measurement")
ax.spines[["top", "right"]].set_visible(False)
plt.tight_layout()

6. Grouped box plots (two categorical variables)

To compare a condition across timepoints, doses, or sites, use hue in seaborn. This produces side-by-side boxes within each x-axis group and is far cleaner than juggling positions manually in matplotlib.

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

rng = np.random.default_rng(42)
n = 40
df = pd.DataFrame({
    "value": np.concatenate([
        rng.normal(50, 6, n), rng.normal(58, 6, n),   # Day 1
        rng.normal(54, 6, n), rng.normal(70, 6, n),   # Day 7
    ]),
    "timepoint": np.repeat(["Day 1", "Day 1", "Day 7", "Day 7"], n),
    "condition": np.tile(np.repeat(["Control", "Treated"], n), 2),
})

fig, ax = plt.subplots(figsize=(6.6, 4.6))
sns.boxplot(data=df, x="timepoint", y="value", hue="condition",
            palette=["#6b7280", "#7c3aed"], ax=ax)
ax.set(xlabel="", ylabel="Expression")
ax.legend(title="", frameon=False)
sns.despine()
plt.tight_layout()

Box plot anatomy: what each line means

7. Export at journal quality

A box plot is only publication-ready once it is exported correctly. Save at 300 DPI minimum, trim whitespace, and use a vector format (PDF or SVG) when the journal accepts it. See our high-resolution export guide for the full checklist.

import matplotlib.pyplot as plt

# ... build your box plot on the axes "ax" ...

fig.savefig(
    "boxplot_figure.png",
    dpi=300,                 # 300 DPI minimum for most journals
    bbox_inches="tight",     # trim surrounding whitespace
    facecolor="white",       # avoid transparent backgrounds in print
)
fig.savefig("boxplot_figure.pdf", bbox_inches="tight")  # vector for line art

Common mistakes and how to fix them

• Boxes will not fill with color — you forgot patch_artist=True in matplotlib.
• Outliers appear twice — when overlaying a strip plot, set fliersize=0 (seaborn) or showfliers=False (matplotlib) on the box plot.
• Reader assumes the line is the mean — state explicitly in the caption that the center line is the median, or add a marker for the mean with showmeans=True.
• Box plot on tiny samples — with n < 10, quartiles are unstable. Overlay the raw points or use a dot plot instead.
• labels vs tick_labels error — newer matplotlib renamed labels to tick_labels; use whichever your version expects.

When to use a box plot vs an alternative

Box plots summarize distribution shape compactly, but they hide multimodality. If the shape of the distribution matters — for example two hidden subpopulations — a violin plot reveals it where a box plot would not. If you mainly need to compare group means with uncertainty, a bar chart with error bars may communicate faster.

Frequently asked questions