diff --git a/r/R/duckdb.R b/r/R/duckdb.R
index a2bf62de2fde2..65c70243e7ab3 100644
--- a/r/R/duckdb.R
+++ b/r/R/duckdb.R
@@ -137,7 +137,13 @@ duckdb_disconnector <- function(con, tbl_name) {
 
 #' Create an Arrow object from a DuckDB connection
 #'
-#' This can be used in pipelines that pass data back and forth between Arrow and DuckDB
+#' This can be used in pipelines that pass data back and forth between Arrow and
+#' DuckDB.
+#'
+#' Note that you can only call `collect()` or `compute()` on the result of this
+#' function once. To work around this limitation, you should either only call
+#' `collect()` as the final step in a pipeline or call `as_arrow_table()` on the
+#' result to materialize the entire Table in-memory.
 #'
 #' @param .data the object to be converted
 #' @return A `RecordBatchReader`.
diff --git a/r/man/to_arrow.Rd b/r/man/to_arrow.Rd
index aed40609a5161..87b8fea36eeda 100644
--- a/r/man/to_arrow.Rd
+++ b/r/man/to_arrow.Rd
@@ -13,7 +13,14 @@ to_arrow(.data)
 A \code{RecordBatchReader}.
 }
 \description{
-This can be used in pipelines that pass data back and forth between Arrow and DuckDB
+This can be used in pipelines that pass data back and forth between Arrow and
+DuckDB.
+}
+\details{
+Note that you can only call \code{collect()} or \code{compute()} on the result of this
+function once. To work around this limitation, you should either only call
+\code{collect()} as the final step in a pipeline or call \code{as_arrow_table()} on the
+result to materialize the entire Table in-memory.
 }
 \examples{
 \dontshow{if (getFromNamespace("run_duckdb_examples", "arrow")()) (if (getRversion() >= "3.4") withAutoprint else force)(\{ # examplesIf}