R r‑statistics.co

tidyr unite() in R: Combine Columns Into One

The unite() function in tidyr combines multiple columns into a single column with a separator. It is the inverse of separate() and the most direct way to build composite identifiers like dates or full names.

By Selva Prabhakaran  ·  Published May 11, 2026  ·  Last updated May 11, 2026
⚡ Quick Answer
unite(df, "full_name", first, last, sep = " ")              # combine 2 cols
unite(df, "id", year, month, day, sep = "-")                # 3 cols with dash
unite(df, "label", year, month, sep = "_", remove = FALSE)  # keep originals
unite(df, "date", -id, sep = "-")                           # all except id
unite(df, "tag", a, b, sep = "_", na.rm = TRUE)             # skip NAs
df |> mutate(full = paste(first, last, sep = " "))          # base R alternative

Need explanation? Read on for examples and pitfalls.

📊 Is unite() the right tool?
STARTcombine 2+ columns into oneunite()combine but keep originalsunite(remove = FALSE)skip NA valuesunite(na.rm = TRUE)combine with pastemutate(new = paste(a, b, sep = ...))separate one into manyseparate()split by regexseparate_wider_regex()glue-style templateglue::glue()

What unite() does in one sentence

unite() takes multiple columns and concatenates their values into a single new column with a separator. By default, the original columns are removed; set remove = FALSE to keep them.

This is the simplest tool for building composite keys (e.g., date from y/m/d), full names, or any concatenated identifier.

Syntax

unite(data, col, ..., sep, remove, na.rm). col is the new column name; ... lists columns to combine; sep is the separator string.

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.
RCombine first and last name
library(tidyr) library(dplyr) library(tibble) df <- tibble::tribble( ~first, ~last, ~age, "Alice", "Smith", 30, "Bob", "Jones", 25 ) df |> unite("full_name", first, last, sep = " ") #> # A tibble: 2 x 2 #> full_name age #> <chr> <dbl> #> 1 Alice Smith 30 #> 2 Bob Jones 25

  

  





Tip
Default sep = "_" is rarely what you want for human-readable text. Always specify sep explicitly: " " for names, "-" for dates, etc. The default underscore is more for machine identifiers.



Five common patterns



1. Two columns into one




  

    
RFirst + last to full

    
df |> unite("full", first, last, sep = " ")

    

      
      
    

    

  

  





The simplest case: two columns become one with a space.



2. Keep originals with remove = FALSE




  

    
RAdd a derived column, keep components

    
df |> unite("full", first, last, sep = " ", remove = FALSE)
#> # A tibble: 2 x 4
#>   full         first last    age
#>   <chr>        <chr> <chr> <dbl>
#> 1 Alice Smith  Alice Smith    30
#> 2 Bob Jones    Bob   Jones    25

    

      
      
    

    

  

  





remove = FALSE adds the new column without dropping the originals.



3. Build a date from components




  

    
RYear, month, day to ISO date

    
df_date <- tibble::tribble(
  ~y,    ~m, ~d,
  2024,  1,  15,
  2024,  3,  20
)

df_date |> unite("date", y, m, d, sep = "-")
#> # A tibble: 2 x 1
#>   date
#>   <chr>
#> 1 2024-1-15
#> 2 2024-3-20

    

      
      
    

    

  

  





For zero-padded months/days, format first: mutate(m = sprintf("%02d", m), d = sprintf("%02d", d)).



4. Skip NAs




  

    
Rna.rm = TRUE removes NA pieces

    
df_na <- tibble::tribble(
  ~a, ~b, ~c,
  "x", "y", "z",
  "x", NA, "z"
)

df_na |> unite("combo", a, b, c, sep = "-", na.rm = TRUE)
#> # A tibble: 2 x 1
#>   combo
#>   <chr>
#> 1 x-y-z
#> 2 x-z

    

      
      
    

    

  

  





na.rm = TRUE skips NA values, joining only the non-missing pieces.



5. Tidyselect for many columns




  

    
RCombine all columns except id

    
df_multi <- tibble::tibble(
  id = 1:2, a = c("x","y"), b = c("p","q"), c = c("1","2")
)

df_multi |> unite("combo", -id, sep = "_")
#> # A tibble: 2 x 2
#>      id combo
#>   <int> <chr>
#> 1     1 x_p_1
#> 2     2 y_q_2

    

      
      
    

    

  

  





-id selects all columns except id. Tidyselect helpers like starts_with() work too.



Key Insight
unite() is just paste() with a clean dplyr-friendly interface. The base R equivalent is mutate(new = paste(a, b, c, sep = "_")). Use unite when working in tidyr/dplyr pipelines; use paste when not.



Common pitfalls



Pitfall 1: forgetting remove = FALSE. Default removes the original columns. If you want to keep them, set remove = FALSE explicitly.



Pitfall 2: numeric inputs converted to strings. unite() always returns a character column. Numeric inputs (like year) become "2024", which is a string. Convert back if needed: mutate(date = as.Date(date)).



Warning
na.rm = FALSE (default) propagates "NA" as a string in the result. With NA in any input column, the output is something like "Alice-NA-Smith". Always set na.rm = TRUE if NAs should be skipped, or fill NAs before unite.



Try it yourself





Try it: Combine iris$Species and a new "year" column (2024) into a single "tag" column like "setosa_2024". Save to ex_tagged.



  

    
RYour turn: build a tag column

    
ex_tagged <- iris |>
  dplyr::mutate(year = 2024) |>
  # your code here

head(ex_tagged, 2)
#> Expected: tag column showing "setosa_2024"

    

      
      
    

    

  

  





Click to reveal solution



  

    
RSolution

    
ex_tagged <- iris |>
  dplyr::mutate(year = 2024) |>
  tidyr::unite("tag", Species, year, sep = "_", remove = FALSE)

head(ex_tagged, 2)
#>   Sepal.Length Sepal.Width Petal.Length Petal.Width tag         Species year
#> 1          5.1         3.5          1.4         0.2 setosa_2024 setosa  2024
#> 2          4.9         3.0          1.4         0.2 setosa_2024 setosa  2024

    

      
      
    

    

  

  





Explanation: unite("tag", Species, year, sep = "_") combines the Species and year columns into "tag". remove = FALSE keeps the originals. Each row's tag is the species name plus year, joined with an underscore.










After mastering unite, look at:



    


  • separate(): split one column into many (inverse)
    • 


  • paste(), paste0(): base R column concatenation
    • 


  • glue::glue(): template-based string interpolation
    • 


  • stringr::str_c(): stringr's vectorized string concatenation
    • 


  • unite() plus arrange(): build a sortable composite key
    • 





For complex composite IDs, glue("{first}_{year}_{country}") inside mutate() is more flexible than unite().



FAQ



How do I combine multiple columns into one in R?



Use tidyr::unite("new", col1, col2, sep = "_") for tidyr style. Or base R: mutate(new = paste(col1, col2, sep = "_")). Both produce a single character column from multiple inputs.



What is the difference between unite and paste in R?



unite() is a dplyr-friendly verb that adds a new column to a data frame and optionally removes the originals. paste() is a base R function that concatenates strings; you wrap it in mutate() to add to a data frame.



How do I keep the original columns after unite?



Set remove = FALSE: unite(df, "new", a, b, sep = "_", remove = FALSE). Default remove = TRUE drops the columns being combined.



How do I handle NA values in unite?



Set na.rm = TRUE: unite("new", a, b, sep = "_", na.rm = TRUE) skips NA values when concatenating. Default na.rm = FALSE includes "NA" as a string in the result.



What separator should I use for unite?



Use a separator that does NOT appear in your data. For names: " ". For dates: "-". For machine identifiers: "_". Avoid characters used in your data (e.g., "," if values contain commas).



        


        
      


      

        

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

        

          About ·
          RSS