diff --git a/src/v/config/configuration.cc b/src/v/config/configuration.cc
index 5b9a461d2bc2..f5cdfef3e6f1 100644
--- a/src/v/config/configuration.cc
+++ b/src/v/config/configuration.cc
@@ -517,6 +517,12 @@ configuration::configuration()
       "Maximum number of bytes returned in fetch request",
       {.needs_restart = needs_restart::no, .visibility = visibility::user},
       55_MiB)
+  , use_fetch_scheduler_group(
+      *this,
+      "use_fetch_scheduler_group",
+      "Use a separate scheduler group for fetch processing",
+      {.needs_restart = needs_restart::no, .visibility = visibility::tunable},
+      true)
   , metadata_status_wait_timeout_ms(
       *this,
       "metadata_status_wait_timeout_ms",
diff --git a/src/v/config/configuration.h b/src/v/config/configuration.h
index d8005941a050..31c21906606f 100644
--- a/src/v/config/configuration.h
+++ b/src/v/config/configuration.h
@@ -120,6 +120,7 @@ struct configuration final : public config_store {
     enum_property<model::timestamp_type> log_message_timestamp_type;
     enum_property<model::compression> log_compression_type;
     property<size_t> fetch_max_bytes;
+    property<bool> use_fetch_scheduler_group;
     property<std::chrono::milliseconds> metadata_status_wait_timeout_ms;
     bounded_property<std::optional<int64_t>> kafka_connection_rate_limit;
     property<std::vector<ss::sstring>> kafka_connection_rate_limit_overrides;
diff --git a/src/v/kafka/server/server.cc b/src/v/kafka/server/server.cc
index a37da6568c2a..c130676ddfa4 100644
--- a/src/v/kafka/server/server.cc
+++ b/src/v/kafka/server/server.cc
@@ -136,6 +136,12 @@ server::server(
     _probe.setup_public_metrics();
 }
 
+ss::scheduling_group server::fetch_scheduling_group() const {
+    return config::shard_local_cfg().use_fetch_scheduler_group()
+             ? _fetch_scheduling_group
+             : ss::default_scheduling_group();
+}
+
 coordinator_ntp_mapper& server::coordinator_mapper() {
     return _group_router.local().coordinator_mapper().local();
 }
diff --git a/src/v/kafka/server/server.h b/src/v/kafka/server/server.h
index d97bc0eca10d..e36c2cdf828e 100644
--- a/src/v/kafka/server/server.h
+++ b/src/v/kafka/server/server.h
@@ -73,9 +73,16 @@ class server final : public net::server {
     ss::future<> apply(ss::lw_shared_ptr<net::connection>) final;
 
     ss::smp_service_group smp_group() const { return _smp_group; }
-    ss::scheduling_group fetch_scheduling_group() const {
-        return _fetch_scheduling_group;
-    }
+
+    /**
+     * @brief Return the scheduling group to use for fetch requests.
+     *
+     * By default, fetches use a dedicated scheduling group, but this may
+     * be changed by configuration to restore the old behavior of lumping
+     * them with most other tasks in the default scheduling group.
+     */
+    ss::scheduling_group fetch_scheduling_group() const;
+
     cluster::topics_frontend& topics_frontend() {
         return _topics_frontend.local();
     }