diff --git a/R/pagination_utils.R b/R/pagination_utils.R
index c53a1f5..28565e5 100644
--- a/R/pagination_utils.R
+++ b/R/pagination_utils.R
@@ -18,6 +18,10 @@
 #'   Providing a value (e.g., `"|"`, `""`, or `"—"`) automatically enables filling and uses that
 #'   value for all cells in the empty rows. This helps maintain consistent vertical positioning
 #'   across all pages. The target row count is the maximum page size across all pages.
+#'   Use `NA` to fill with missing values while preserving original column types.
+#'   If you need consistent types across all pages, use `NA`.
+#'   Because `fill_empty` is a character placeholder, pages are coerced to character
+#'   columns when filling is applied (except when `fill_empty = NA`).
 #'
 #' @return A list of data frames, one for each page. When `split_by` is used, the list
 #'   is named with the group values. If a group spans multiple pages (when combined with
@@ -177,8 +181,13 @@ paginate_table <- function(
     }
   }
 
-  if (!is.null(fill_empty) && (!is.character(fill_empty) || length(fill_empty) != 1)) {
-    stop("`fill_empty` must be NULL or a single character string.")
+  if (!is.null(fill_empty)) {
+    fill_is_single_na <- length(fill_empty) == 1 && isTRUE(is.na(fill_empty))
+    fill_is_single_character <- is.character(fill_empty) && length(fill_empty) == 1
+
+    if (!(fill_is_single_na || fill_is_single_character)) {
+      stop("`fill_empty` must be NULL, NA, or a single character string.")
+    }
   }
 
     # Split data based on method
@@ -231,12 +240,28 @@ paginate_table <- function(
       if (page_nrows < rows_per_page) {
         rows_difference <- rows_per_page - page_nrows
 
-        # Create empty rows with specified fill value
-        empty_df <- data.frame(
-          matrix(fill_empty, nrow = rows_difference, ncol = ncol(data)),
-          stringsAsFactors = FALSE
-        )
-        colnames(empty_df) <- colnames(data)
+        if (isTRUE(is.na(fill_empty))) {
+          # Preserve original column classes when filling with missing values.
+          empty_df <- as.data.frame(
+            lapply(page, function(col) rep(col[NA_integer_], rows_difference)),
+            stringsAsFactors = FALSE,
+            check.names = FALSE
+          )
+        } else {
+          # Character placeholder fill uses character columns across all fields.
+          page <- as.data.frame(
+            lapply(page, as.character),
+            stringsAsFactors = FALSE,
+            check.names = FALSE
+          )
+
+          empty_df <- data.frame(
+            matrix(fill_empty, nrow = rows_difference, ncol = ncol(data)),
+            stringsAsFactors = FALSE,
+            check.names = FALSE
+          )
+          colnames(empty_df) <- colnames(data)
+        }
 
         # Append to page
         page <- rbind(page, empty_df)
diff --git a/man/paginate_table.Rd b/man/paginate_table.Rd
index 04ffefe..d165a2f 100644
--- a/man/paginate_table.Rd
+++ b/man/paginate_table.Rd
@@ -23,7 +23,11 @@ At least one of \code{rows_per_page} or \code{split_by} must be provided.}
 with empty rows to match the target row count. Default is \code{NULL} (no filling).
 Providing a value (e.g., \code{"|"}, \code{""}, or \code{"—"}) automatically enables filling and uses that
 value for all cells in the empty rows. This helps maintain consistent vertical positioning
-across all pages. The target row count is the maximum page size across all pages.}
+across all pages. The target row count is the maximum page size across all pages.
+Use \code{NA} to fill with missing values while preserving original column types.
+If you need consistent types across all pages, use \code{NA}.
+Because \code{fill_empty} is a character placeholder, pages are coerced to character
+columns when filling is applied (except when \code{fill_empty = NA}).}
 }
 \value{
 A list of data frames, one for each page. When \code{split_by} is used, the list
diff --git a/tests/testthat/test_paginate_table.R b/tests/testthat/test_paginate_table.R
index 1249ec2..9b65a6a 100644
--- a/tests/testthat/test_paginate_table.R
+++ b/tests/testthat/test_paginate_table.R
@@ -60,12 +60,12 @@ test_that("paginate_table validates inputs correctly", {
 
   expect_error(
     paginate_table(mtcars, 10, fill_empty = TRUE),
-    "`fill_empty` must be NULL or a single character string"
+    "`fill_empty` must be NULL, NA, or a single character string"
   )
 
   expect_error(
     paginate_table(mtcars, 10, fill_empty = c("|", "-")),
-    "`fill_empty` must be NULL or a single character string"
+    "`fill_empty` must be NULL, NA, or a single character string"
   )
 
   expect_error(
@@ -198,3 +198,34 @@ test_that("paginate_table split_by handles single group", {
   expect_equal(length(pages), 1)
   expect_equal(nrow(pages[[1]]), nrow(df_single))
 })
+
+test_that("paginate_table fill_empty works with Date columns", {
+  df <- data.frame(
+    aval = runif(40, 10, 50),
+    dt = as.Date(rep(c("2025-01-02", "2025-02-03", "2025-03-04", "2025-04-05"), 10))
+  )
+
+  expect_no_error({
+    pages <- paginate_table(data = df, rows_per_page = 7, fill_empty = " ")
+    expect_true(length(pages) > 0)
+    expect_equal(nrow(pages[[length(pages)]]), 7)
+    expect_type(pages[[length(pages)]]$dt, "character")
+    expect_equal(pages[[length(pages)]][7, "dt"], " ")
+  })
+})
+
+test_that("paginate_table supports fill_empty = NA preserving column types", {
+  df <- data.frame(
+    aval = runif(40, 10, 50),
+    dt = as.Date(rep(c("2025-01-02", "2025-02-03", "2025-03-04", "2025-04-05"), 10))
+  )
+
+  pages <- paginate_table(data = df, rows_per_page = 7, fill_empty = NA)
+  last_page <- pages[[length(pages)]]
+
+  expect_equal(nrow(last_page), 7)
+  expect_type(last_page$aval, "double")
+  expect_s3_class(last_page$dt, "Date")
+  expect_true(anyNA(last_page[7, "aval"]))
+  expect_true(anyNA(last_page[7, "dt"]))
+})