diff --git a/R/pkg/R/DataFrame.R b/R/pkg/R/DataFrame.R
index 176bb3b8a8d0c..aaa3349d57506 100644
--- a/R/pkg/R/DataFrame.R
+++ b/R/pkg/R/DataFrame.R
@@ -1191,6 +1191,9 @@ setMethod("collect",
                     vec <- do.call(c, col)
                     stopifnot(class(vec) != "list")
                     class(vec) <- PRIMITIVE_TYPES[[colType]]
+                    if (is.character(vec) && stringsAsFactors) {
+                      vec <- as.factor(vec)
+                    }
                     df[[colIndex]] <- vec
                   } else {
                     df[[colIndex]] <- col
diff --git a/R/pkg/tests/fulltests/test_sparkSQL.R b/R/pkg/tests/fulltests/test_sparkSQL.R
index 4382ef2ed4525..0c8118a7c73f3 100644
--- a/R/pkg/tests/fulltests/test_sparkSQL.R
+++ b/R/pkg/tests/fulltests/test_sparkSQL.R
@@ -499,6 +499,12 @@ test_that("create DataFrame with different data types", {
   expect_equal(collect(df), data.frame(l, stringsAsFactors = FALSE))
 })
 
+test_that("SPARK-17902: collect() with stringsAsFactors enabled", {
+  df <- suppressWarnings(collect(createDataFrame(iris), stringsAsFactors = TRUE))
+  expect_equal(class(iris$Species), class(df$Species))
+  expect_equal(iris$Species, df$Species)
+})
+
 test_that("SPARK-17811: can create DataFrame containing NA as date and time", {
   df <- data.frame(
     id = 1:2,