dplyr first() in R: Get the First Value of a Vector

The first() function in dplyr returns the FIRST element of a vector, with optional default for empty input and order_by for sorting. It is the cleaner alternative to x[1] inside dplyr pipelines.

By Selva Prabhakaran · Published May 12, 2026 · Last updated May 12, 2026

⚡ Quick Answer

first(c(10, 20, 30))                  # 10
first(c())                             # default = NA
first(x, default = 0)                  # specify fallback
first(x, order_by = ts)                # first by ts column
df |> summarise(first_val = first(value))
df |> group_by(g) |> summarise(first_val = first(value))

Need explanation? Read on for examples and pitfalls.

📊 Is first() the right tool?

What first() does in one sentence

first(x, default = NA, order_by = NULL) returns the first element of x; on empty input it returns default instead of erroring. With order_by, x is reordered by that vector before extracting position 1.

The dplyr-friendly version of x[1]. Returns a scalar, integrates with summarise / mutate, and handles the empty-input edge case gracefully.

Syntax

first(x, default = NA, order_by = NULL). Default is NA of x's type.

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.

RFirst value of a vector

library(dplyr) first(c(10, 20, 30)) #> [1] 10 first(integer(0)) #> [1] NA

Tip

first() returns a SCALAR; slice_head() returns a tibble. Use first() inside summarise / mutate where you want a single value; use slice_head for row-level filtering.

Five common patterns

1. First element

RPosition 1

first(c(10, 20, 30)) #> [1] 10

2. With a custom default

RWhat if vector is empty?

first(integer(0), default = 0) #> [1] 0 first(c(), default = "unknown") #> [1] "unknown"

3. Per-group first

RFirst value per group inside summarise

df <- data.frame( user = c("a","a","a","b","b"), ts = c(3, 1, 2, 5, 4), val = c(10, 20, 30, 40, 50) ) df |> group_by(user) |> summarise(first_val = first(val)) #> # A tibble: 2 x 2 #> user first_val #> a 10 #> b 40

By default, "first" means the first row order. To get chronological first, use order_by.

4. First by another column's order

REarliest by timestamp per user

df |> group_by(user) |> summarise(first_val = first(val, order_by = ts)) #> # A tibble: 2 x 2 #> user first_val #> a 20 <-- val at ts=1 #> b 50 <-- val at ts=4

order_by = ts sorts by ts before picking position 1.

5. First non-NA

RSkip leading NAs

x <- c(NA, NA, 30, NA, 50) first(na.omit(x)) #> [1] 30 # Or via coalesce: coalesce(x[1], x[2], x[3], x[4], x[5]) #> [1] 30

first(na.omit(x)) returns the first non-NA value.

Key Insight

first() and last() are scalar-returning equivalents of slice_head(1) and slice_tail(1). Use first/last inside summarise; use slice_head/slice_tail when you want the row as a tibble. Different return shapes, similar intent.

first() vs head() vs slice_head() vs nth()

Four ways to get "first" elements in R, with different return shapes.

Function	Input	Output	Best for
`first(x)`	Vector	Scalar	dplyr summarise/mutate
`head(x, 1)`	Vector	Length-1 vector	Quick base R
`head(df, 1)`	Data frame	1-row df	First row of a frame
`slice_head(df, n = 1)`	Data frame	1-row tibble	dplyr; group-aware
`nth(x, 1)`	Vector	Scalar	Pick by index, generalizes

When to use which:

first(x) for scalar output inside summarise.
slice_head for row-level extraction.
nth(x, k) for arbitrary positions.
head for quick base R.

A practical workflow

Use first() for the "first event per user" pattern alongside group_by.

RFirst event chronologically per user

df |> group_by(user) |> summarise( first_visit = first(timestamp, order_by = timestamp), first_action = first(action, order_by = timestamp), .groups = "drop" )

Per user, get the chronologically first visit and action. order_by = timestamp ensures correct order regardless of input row order.

Common pitfalls

Pitfall 1: confusing first() with slice_head(). first returns a scalar; slice_head returns a tibble. Different shapes. Inside summarise, use first; inside mutate to keep multiple rows, use slice_head.

Pitfall 2: order_by silent if data is unsorted. Without order_by, first uses physical row order. If your data isn't sorted, "first" is whatever happened to be loaded first.

Warning

first() returns NA on empty input by default. If your downstream code can't handle NA, set default explicitly: first(x, default = 0).

Why first/last beat x[1] inside dplyr

Inside summarise() and mutate(), first(x) is preferred over x[1] for two reasons. First, it handles the empty-input case gracefully: first(integer(0)) returns NA instead of erroring; integer(0)[1] returns NA but with a warning in some configurations. Second, it accepts an order_by argument, so you can express "first chronologically" without an explicit arrange step. For pure speed, x[1] is slightly faster, but the readability and safety wins of first() make it the right default in pipeline code. For interactive sketching x[1] is fine.

Try it yourself

Try it: For each cyl group in mtcars, get the mpg of the first car (in row order). Save to ex_first.

RYour turn: first mpg per cyl

ex_first <- mtcars |> # your code here ex_first #> Expected: 3 rows (one per cyl) with first_mpg

Click to reveal solution

RSolution

ex_first <- mtcars |> group_by(cyl) |> summarise(first_mpg = first(mpg)) ex_first #> # A tibble: 3 x 2 #> cyl first_mpg #> 4 22.8 #> 6 21.0 #> 8 18.7

Explanation: first(mpg) per cyl group returns the mpg of the first row in each group (in original row order).

After mastering first, look at:

last(): last value (mirror)
nth(x, k): arbitrary position
slice_head() / slice_tail(): row-level versions
head() / tail(): base R; data-frame friendly
coalesce(): first non-NA across vectors
na.omit(): drop NAs before first()

For "first non-NA", first(na.omit(x)) is the cleanest pattern.

FAQ

What does first do in dplyr?

first(x) returns the first element of a vector as a scalar. Used inside summarise / mutate, with optional default for empty input and order_by for sorting.

What is the difference between first() and head() in R?

first(x) returns a scalar (single value). head(x, 1) returns a length-1 vector. They differ in shape, not semantics. For data frames, head(df, 1) returns a 1-row data frame; first() doesn't apply directly.

How do I get the first non-NA value in dplyr?

first(na.omit(x)) drops NAs first, then takes position 1. Or coalesce(x[1], x[2], ...) for explicit fallback.

How do I use order_by in first()?

first(x, order_by = ts) sorts x by the corresponding ts values, then takes the first. Useful for "earliest event" queries when input rows aren't sorted.

How do I get the first value per group?

df |> group_by(g) |> summarise(first_val = first(x)). Add order_by = ts to specify sort order within group.

Navigate

Tidyverse packages

Deep dives

Wrangling & EDA

Statistics

Machine Learning

Time Series

By Industry

Reporting & Apps

Levels

dplyr first() in R: Get the First Value of a Vector

What first() does in one sentence

Syntax

Five common patterns

1. First element

2. With a custom default

3. Per-group first

4. First by another column's order

5. First non-NA

first() vs head() vs slice_head() vs nth()

A practical workflow

Common pitfalls

Why first/last beat x[1] inside dplyr

Try it yourself

FAQ

Navigate

Tidyverse packages

Deep dives

Wrangling & EDA

Statistics

Machine Learning

Time Series

By Industry

Reporting & Apps

Levels

dplyr first() in R: Get the First Value of a Vector

What first() does in one sentence

Syntax

Five common patterns

1. First element

2. With a custom default

3. Per-group first

4. First by another column's order

5. First non-NA

first() vs head() vs slice_head() vs nth()

A practical workflow

Common pitfalls

Why first/last beat x[1] inside dplyr

Try it yourself

Related dplyr functions

FAQ