diff --git a/NEWS.md b/NEWS.md
index e8cea03b..d7d1c3b1 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -4,6 +4,9 @@
 
 * `as_factor()` and `fct_inorder()` accept NA levels (#98).
 
+* `fct_lump()` gains `weights` argument (#70, @wilkox) to weight value
+  frequencies before lumping them together (#68).
+
 # forcats 0.2.0
 
 ## New functions
diff --git a/R/lump.R b/R/lump.R
index ddf953b5..e36e1392 100644
--- a/R/lump.R
+++ b/R/lump.R
@@ -14,6 +14,8 @@
 #'   Positive `prop` preserves values that appear at least
 #'   `prop` of the time. Negative `prop` preserves values that
 #'   appear at most `-prop` of the time.
+#' @param weights An optional numeric vector giving weights for frequency of
+#'   each value (not level) in f.
 #' @param other_level Value of level used for "other" values. Always
 #'   placed at end of levels.
 #' @param ties.method A character string specifying how ties are
@@ -39,20 +41,36 @@
 #' fct_lump(x, n = -3)
 #' fct_lump(x, prop = -0.1)
 #'
+#' # Use weighted frequencies
+#' w <- c(rep(2, 50), rep(1, 50))
+#' fct_lump(x, n = 5, weights = w)
+#'
 #' # Use ties.method to control how tied factors are collapsed
 #' fct_lump(x, n = 6)
 #' fct_lump(x, n = 6, ties.method = "max")
 #'
-fct_lump <- function(f, n, prop, other_level = "Other",
+fct_lump <- function(f, n, prop, weights = NULL, other_level = "Other",
                      ties.method = c("min", "average", "first", "last", "random", "max")) {
   f <- check_factor(f)
   ties.method <- match.arg(ties.method)
 
+  if (!is.null(weights)) {
+    if (!is.numeric(weights)) {
+      stop("`weights` must be a numeric vector", call. = FALSE)
+    } else if (length(f) != length(weights)) {
+      stop("Different lengths of `f` and `weights`", call. = FALSE)
+    }
+  }
+
   levels <- levels(f)
-  count <- table(f)
+  if (is.null(weights)) {
+    count <- table(f)
+  } else {
+    count <- tapply(weights, f, FUN = sum)
+  }
 
   if (!xor(missing(n), missing(prop))) {
-    new_levels <- ifelse(!in_smallest(table(f)), levels, other_level)
+    new_levels <- ifelse(!in_smallest(count), levels, other_level)
   } else if (!missing(n)) {
     if (n < 0) {
       rank <- rank(count, ties = ties.method)
diff --git a/man/fct_lump.Rd b/man/fct_lump.Rd
index 70cd333e..7835e2ac 100644
--- a/man/fct_lump.Rd
+++ b/man/fct_lump.Rd
@@ -4,8 +4,8 @@
 \alias{fct_lump}
 \title{Lump together least/most common factor levels into "other"}
 \usage{
-fct_lump(f, n, prop, other_level = "Other", ties.method = c("min",
-  "average", "first", "last", "random", "max"))
+fct_lump(f, n, prop, weights = NULL, other_level = "Other",
+  ties.method = c("min", "average", "first", "last", "random", "max"))
 }
 \arguments{
 \item{f}{A factor.}
@@ -23,6 +23,9 @@ Positive \code{prop} preserves values that appear at least
 \code{prop} of the time. Negative \code{prop} preserves values that
 appear at most \code{-prop} of the time.}
 
+\item{weights}{An optional numeric vector giving weights for frequency of
+each value (not level) in f.}
+
 \item{other_level}{Value of level used for "other" values. Always
 placed at end of levels.}
 
@@ -51,6 +54,10 @@ fct_lump(x, prop = 0.1)
 fct_lump(x, n = -3)
 fct_lump(x, prop = -0.1)
 
+# Use weighted frequencies
+w <- c(rep(2, 50), rep(1, 50))
+fct_lump(x, n = 5, weights = w)
+
 # Use ties.method to control how tied factors are collapsed
 fct_lump(x, n = 6)
 fct_lump(x, n = 6, ties.method = "max")
diff --git a/tests/testthat/test-fct_lump.R b/tests/testthat/test-fct_lump.R
index 576dcade..c234923a 100644
--- a/tests/testthat/test-fct_lump.R
+++ b/tests/testthat/test-fct_lump.R
@@ -55,6 +55,63 @@ test_that("different behaviour when apply tie function", {
   }
 })
 
+test_that("bad weights generate error messages", {
+  f <- c("a", "a", "a", "b", "b", "c", "d", "e", "f", "g")
+
+  w <- c(
+    "one",
+    "three",
+    "four",
+    "five",
+    "nine",
+    "twenty",
+    "three",
+    "seven",
+    "twelve",
+    "two"
+  )
+  expect_error(fct_lump(f, n = 2, weight = w), "must be a numeric vector")
+
+  w <- c(1, 2, 3, 4, 5, 6, 7, 8, 9)
+  expect_error(fct_lump(f, n = 4, weight = w), "Different lengths")
+})
+
+test_that("values are correctly weighted", {
+  f <- c("a", "a", "a", "b", "b", "c", "d", "e", "f", "g")
+  w <- c( 0.2, 0.2, 0.6, 2,   2,   6,   4,   2,   2,   1)
+  f2 <- c(
+    "a",
+    rep("b", 4),
+    rep("c", 6),
+    rep("d", 4),
+    rep("e", 2),
+    rep("f", 2),
+    "g"
+  )
+
+  expect_equal(levels(fct_lump(f, weights = w)), levels(fct_lump(f2)))
+  expect_equal(
+    levels(fct_lump(f, n = 1, weights = w)),
+    levels(fct_lump(f2, n = 1))
+  )
+  expect_equal(
+    levels(fct_lump(f, n = -2, weights = w, ties.method = "first")),
+    levels(fct_lump(f2, n = -2, ties.method = "first"))
+  )
+  expect_equal(
+    levels(fct_lump(f, n = 99, weights = w)),
+    levels(fct_lump(f2, n = 99))
+  )
+  expect_equal(
+    levels(fct_lump(f, prop = 0.01, weights = w)),
+    levels(fct_lump(f2, prop = 0.01))
+  )
+  expect_equal(
+    levels(fct_lump(f, prop = -0.25, weights = w, ties.method = "max")),
+    levels(fct_lump(f2, prop = -0.25, ties.method = "max"))
+  )
+})
+
 # Default -----------------------------------------------------------------
 
 test_that("lumps smallest", {