R r‑statistics.co

ggplot2 geom_histogram() in R: Histograms With Examples

The geom_histogram() function in ggplot2 bins a continuous variable into intervals and draws bars showing the count per bin. Use binwidth or bins to control bin size; fill to color by group.

By Selva Prabhakaran  ·  Published May 11, 2026  ·  Last updated May 11, 2026
⚡ Quick Answer
ggplot(df, aes(x)) + geom_histogram()                          # default 30 bins
ggplot(df, aes(x)) + geom_histogram(binwidth = 0.5)            # fixed bin width
ggplot(df, aes(x)) + geom_histogram(bins = 50)                 # 50 bins total
ggplot(df, aes(x, fill = group)) + geom_histogram(position = "identity", alpha = 0.5)
ggplot(df, aes(x)) + geom_histogram(aes(y = ..density..))      # density scale
ggplot(df, aes(x)) + geom_histogram() + facet_wrap(~ group)    # one panel per group
ggplot(df, aes(x)) + geom_histogram() + geom_density()         # histogram + curve

Need explanation? Read on for examples and pitfalls.

📊 Is geom_histogram() the right tool?
STARTdistribution of one continuousgeom_histogram()compare distributions across groupsaes(fill = grp) + geom_histogram(alpha)smooth shape (no bins)geom_density()distribution by group, separate panels+ facet_wrap(~ grp)discrete (counted) datageom_bar() insteadcumulative distributionstat_ecdf()two continuousgeom_hex() or geom_density_2d()

What geom_histogram() does in one sentence

geom_histogram() divides the x range into bins and draws a bar of height equal to the number of observations in each bin. Default is 30 bins. Specify binwidth to fix bin width in data units, or bins to fix the count.

Histograms are the default tool for understanding the SHAPE of a continuous distribution: skew, multimodality, outliers, range. Use them in EDA before any modeling.

Syntax

geom_histogram() requires only aes(x). The y axis is computed (count by default).

Run live
Run live, no install needed. Every R block on this page runs in your browser. Click Run, edit the code, re-run instantly. No setup.
RLoad ggplot2 and inspect diamonds
library(ggplot2) range(diamonds$carat) #> [1] 0.2 5.01 nrow(diamonds) #> [1] 53940

  

  





The full signature:



geom_histogram(mapping = NULL, data = NULL, stat = "bin", position = "stack",
               ..., binwidth = NULL, bins = NULL, na.rm = FALSE,
               orientation = NA, show.legend = NA, inherit.aes = TRUE)



Tip
Always specify binwidth or bins explicitly. The default (30 bins) is rarely right. ggplot2 issues a console message when you do not set one, hinting at this. Pick binwidth based on the data's units (e.g., 0.5 for carat, 1 for ages, 100 for prices).



Six common patterns



1. Default histogram




  

    
RDefault 30 bins

    
ggplot(diamonds, aes(x = carat)) +
  geom_histogram()
#> `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

    

      
      
    

    

  

  





The console message reminds you the default 30 bins may not suit your data. The plot still renders.



2. Fixed binwidth




  

    
Rbinwidth = 0.1 carat

    
ggplot(diamonds, aes(x = carat)) +
  geom_histogram(binwidth = 0.1, fill = "steelblue", color = "white")

    

      
      
    

    

  

  





binwidth = 0.1 means each bin is 0.1 carats wide. White outlines (color = "white") separate bars visually.



3. Fill by group with overlap




  

    
RCompare two cuts

    
ggplot(diamonds[diamonds$cut %in% c("Ideal", "Fair"), ],
       aes(x = carat, fill = cut)) +
  geom_histogram(binwidth = 0.1, position = "identity", alpha = 0.5)

    

      
      
    

    

  

  





fill = cut colors bars by cut. position = "identity" overlays bars at the same x; alpha = 0.5 makes them semi-transparent so both are visible.



4. Density scale (y = ..density..)




  

    
RUse density instead of count

    
ggplot(diamonds, aes(x = carat, y = ..density..)) +
  geom_histogram(binwidth = 0.1, fill = "steelblue") +
  labs(y = "density")

    

      
      
    

    

  

  





y = ..density.. rescales the y axis so the total area sums to 1. Useful when comparing groups of different sizes (count would mislead).



5. Histogram + density curve




  

    
RHistogram with density overlay

    
ggplot(diamonds, aes(x = carat)) +
  geom_histogram(aes(y = ..density..), binwidth = 0.1,
                 fill = "lightgray", color = "white") +
  geom_density(color = "firebrick", linewidth = 1)

    

      
      
    

    

  

  





The histogram shows binned counts; the density line shows a smooth estimate. Both on the same y scale (density) for fair comparison.



6. Faceted by group




  

    
ROne histogram per cut

    
ggplot(diamonds, aes(x = carat)) +
  geom_histogram(binwidth = 0.1, fill = "steelblue") +
  facet_wrap(~ cut, ncol = 2) +
  theme_minimal()

    

      
      
    

    

  

  





facet_wrap() creates one panel per cut value. Each panel has its own counts, making it easy to compare distribution shapes across groups.



Key Insight
binwidth controls the appearance of the histogram more than any other choice. Too few bins (binwidth too large) hides structure. Too many bins (binwidth too small) shows noise as if it were signal. Try a few values and pick the one that reveals the distribution clearly without being jagged.



geom_histogram() vs base R hist()



Base R hist() is one-shot; ggplot2's geom_histogram() integrates with the rest of the layered grammar.







Task
ggplot2
Base R






Default histogram
geom_histogram()
hist(x)




Specific binwidth
binwidth = 0.5
breaks = seq(0, 5, 0.5)




N bins
bins = 50
breaks = 50




Density y scale
aes(y = ..density..)
freq = FALSE




Compare groups
aes(fill = grp)
hist() per group, manually




Add density curve
+ geom_density()
lines(density(x))







When to use which:



    


  • Use ggplot2 for any multi-group, faceted, or compound histogram.
    • 


  • Use base hist() for one-line interactive checks.
    • 





Common pitfalls



Pitfall 1: not setting binwidth. Default 30 bins rarely matches your data. Always specify either binwidth (fixed width per bin) or bins (target count of bins).



Pitfall 2: stacked bars when overlay was intended. Default position = "stack" stacks bars from different fill groups vertically. To OVERLAY them, set position = "identity" and use alpha = 0.5.



Warning
Bar STACKING in geom_histogram with fill misleads readers. If your goal is comparing distribution shapes across groups, stacking is wrong (the second group's shape gets distorted by the first). Use position = "identity" with alpha, or facet_wrap(), or geom_density() colored by group.



Pitfall 3: y axis is count by default; comparing groups of different sizes is misleading. A group with 100 obs and one with 10,000 obs both show their counts, making the smaller group look flat. Use aes(y = ..density..) for fair comparison.



Try it yourself





Try it: Plot the distribution of mpg$hwy (highway miles per gallon). Use a binwidth of 2, fill the bars steelblue, and add a vertical line at the mean. Save to ex_plot.



  

    
RYour turn: histogram with mean line

    
# Try it: histogram + mean line
ex_plot <- ggplot(mpg, aes(x = hwy)) +
  # your code here

print(ex_plot)
#> Expected: histogram with one vertical red line at mean(hwy)

    

      
      
    

    

  

  





Click to reveal solution



  

    
RSolution

    
ex_plot <- ggplot(mpg, aes(x = hwy)) +
  geom_histogram(binwidth = 2, fill = "steelblue", color = "white") +
  geom_vline(xintercept = mean(mpg$hwy), color = "firebrick",
             linetype = "dashed", linewidth = 1) +
  labs(x = "Highway MPG", y = "Count")

print(ex_plot)

    

      
      
    

    

  

  





Explanation: geom_histogram(binwidth = 2) bins highway mpg into 2-unit intervals. geom_vline(xintercept = mean(...)) adds a vertical reference line at the mean. Dashed red makes it visually distinct from the bars.










After mastering geom_histogram(), look at:



    


  • geom_density(): smooth kernel density estimate (no bins)
    • 


  • geom_freqpoly(): line version of histogram (less visually heavy when comparing groups)
    • 


  • geom_dotplot(): each observation as a stacked dot
    • 


  • stat_bin(): the underlying stat geom_histogram uses
    • 


  • stat_ecdf(): empirical cumulative distribution function plot
    • 


  • geom_violin(): density mirrored around a center for grouped comparisons
    • 





For two-dimensional histograms, see geom_hex() and geom_bin2d().



FAQ



How do I change the number of bins in ggplot2 histogram?



Use bins = N for a target count or binwidth = W for a fixed bin width: geom_histogram(bins = 50) or geom_histogram(binwidth = 0.1). Set ONE; setting both produces a warning.



How do I add a density curve to a ggplot2 histogram?



Set y to density first, then add geom_density(): geom_histogram(aes(y = ..density..)) + geom_density(). Without rescaling y to density, the curve and histogram are on different scales.



How do I overlay two histograms in ggplot2?



Set position = "identity" and use transparency: geom_histogram(aes(fill = group), position = "identity", alpha = 0.5). Default position is "stack" which adds bars on top of each other (rarely what you want).



What is the difference between geom_histogram and geom_bar?



geom_histogram() works on a CONTINUOUS x variable, binning it into intervals. geom_bar() works on a CATEGORICAL x, counting rows per category. Use histogram for numeric distributions, bar for category counts.



How do I plot histogram with percentages?



Compute density and multiply: geom_histogram(aes(y = ..density.. * binwidth_value * 100)). Or use ..count.. / sum(..count..) * 100. Format y axis with scale_y_continuous(labels = scales::percent_format(scale = 1)).



        


        
      


      

        

        
© 2016-2026 Selva Prabhakaran.
          This work is licensed under the Creative Commons License.
        

        

          About ·
          RSS