diff --git a/R/pkg/R/sparkR.R b/R/pkg/R/sparkR.R
index e4a11a5f78..924ebbf37e 100644
--- a/R/pkg/R/sparkR.R
+++ b/R/pkg/R/sparkR.R
@@ -655,6 +655,40 @@ sparkCheckInstall <- function(sparkHome, master, deployMode) {
     } else {
       if (interactive() || isMasterLocal(master)) {
         message("Spark not found in SPARK_HOME: ", sparkHome)
+        # If EXISTING_SPARKR_BACKEND_PORT environment variable is set, assume
+        # that we're in Spark submit. spark-submit always sets Spark home
+        # so this case should not happen. This is just a safeguard.
+        isSparkRSubmit <- Sys.getenv("EXISTING_SPARKR_BACKEND_PORT", "") != ""
+
+        # SPARKR_ASK_INSTALLATION is an internal environment variable in case
+        # users want to disable this behavior. This environment variable should
+        # be removed if no user complains. This environment variable was added
+        # in case other notebook projects are affected.
+        if (!isSparkRSubmit && Sys.getenv("SPARKR_ASK_INSTALLATION", "TRUE") == "TRUE") {
+          # Finally, we're either plain R shell or Rscript.
+          msg <- paste0(
+            "Will you download and install (or reuse if it exists) Spark package ",
+            "under the cache [", sparkCachePath(), "]? (y/n): ")
+
+          answer <- NA
+          while (is.na(answer) || (answer != "y" && answer != "n")) {
+            # Dispatch on R shell in case readLines does not work in RStudio
+            # See https://stackoverflow.com/questions/30191232/use-stdin-from-within-r-studio
+            if (interactive()) {
+              answer <- readline(prompt = msg)
+            } else {
+              cat(msg)
+              answer <- readLines("stdin", n = 1)
+            }
+          }
+          if (answer == "n") {
+            stop(paste0(
+             "Please make sure Spark package is installed in this machine.\n",
+             "  - If there is one, set the path in sparkHome parameter or ",
+             "environment variable SPARK_HOME.\n",
+             "  - If not, you may run install.spark function to do the job."))
+          }
+        }
         packageLocalDir <- install.spark()
         packageLocalDir
       } else if (isClientMode(master) || deployMode == "client") {
diff --git a/docs/sparkr-migration-guide.md b/docs/sparkr-migration-guide.md
index 32836cdac5..3a937b729a 100644
--- a/docs/sparkr-migration-guide.md
+++ b/docs/sparkr-migration-guide.md
@@ -26,6 +26,10 @@ Note that this migration guide describes the items specific to SparkR.
 Many items of SQL migration can be applied when migrating SparkR to higher versions.
 Please refer [Migration Guide: SQL, Datasets and DataFrame](sql-migration-guide.html).
 
+## Upgrading from SparkR 3.1 to 3.2
+
+ - Previously, SparkR automatically downloaded and installed the Spark distribution in user' cache directory to complete SparkR installation when SparkR runs in a plain R shell or Rscript, and the Spark distribution cannot be found. Now, it asks if users want to download and install or not. To restore the previous behavior, set `SPARKR_ASK_INSTALLATION` environment variable to `FALSE`.
+
 ## Upgrading from SparkR 2.4 to 3.0
 
  - The deprecated methods `parquetFile`, `saveAsParquetFile`, `jsonFile`, `jsonRDD` have been removed. Use `read.parquet`, `write.parquet`, `read.json` instead.