diff --git a/conf/scylla.yaml b/conf/scylla.yaml
index 4951be763a..fa4002baf1 100644
--- a/conf/scylla.yaml
+++ b/conf/scylla.yaml
@@ -539,9 +539,6 @@ commitlog_total_space_in_mb: -1
 # Keep at 12 for new clusters.
 murmur3_partitioner_ignore_msb_bits: 12
 
-# Bypass in-memory data cache (the row cache) when performing reversed queries.
-# reversed_reads_auto_bypass_cache: false
-
 # Use on a new, parallel algorithm for performing aggregate queries.
 # Set to `false` to fall-back to the old algorithm.
 # enable_parallelized_aggregation: true
diff --git a/db/config.cc b/db/config.cc
index 1d4b7f310b..3f63e1b290 100644
--- a/db/config.cc
+++ b/db/config.cc
@@ -1066,8 +1066,6 @@ db::config::config(std::shared_ptr<db::extensions> exts)
         "To preserve backwards compatibility on old clusters, Scylla's default setting is `warn`. "
         "New clusters have this option set to `true` by scylla.yaml (which overrides the default `warn`), "
         "to make sure that trying to create an invalid view causes an error.")
-    , reversed_reads_auto_bypass_cache(this, "reversed_reads_auto_bypass_cache", liveness::LiveUpdate, value_status::Used, false,
-            "Bypass in-memory data cache (the row cache) when performing reversed queries.")
     , enable_cql_config_updates(this, "enable_cql_config_updates", liveness::LiveUpdate, value_status::Used, true,
             "Make the system.config table UPDATEable.")
     , enable_parallelized_aggregation(this, "enable_parallelized_aggregation", liveness::LiveUpdate, value_status::Used, true,
diff --git a/db/config.hh b/db/config.hh
index c58ceb9a5f..931fd68483 100644
--- a/db/config.hh
+++ b/db/config.hh
@@ -396,7 +396,6 @@ public:
     named_value<bool> cdc_dont_rewrite_streams;
     named_value<tri_mode_restriction> strict_allow_filtering;
     named_value<tri_mode_restriction> strict_is_not_null_in_views;
-    named_value<bool> reversed_reads_auto_bypass_cache;
     named_value<bool> enable_cql_config_updates;
     named_value<bool> enable_parallelized_aggregation;
 
diff --git a/replica/database.cc b/replica/database.cc
index 7a935c38a8..6b1fa77db5 100644
--- a/replica/database.cc
+++ b/replica/database.cc
@@ -1257,7 +1257,6 @@ keyspace::make_column_family_config(const schema& s, const database& db) const {
     cfg.statement_scheduling_group = _config.statement_scheduling_group;
     cfg.enable_metrics_reporting = db_config.enable_keyspace_column_family_metrics();
     cfg.enable_node_aggregated_table_metrics = db_config.enable_node_aggregated_table_metrics();
-    cfg.reversed_reads_auto_bypass_cache = db_config.reversed_reads_auto_bypass_cache;
     cfg.tombstone_warn_threshold = db_config.tombstone_warn_threshold();
     cfg.view_update_concurrency_semaphore_limit = _config.view_update_concurrency_semaphore_limit;
     cfg.data_listeners = &db.data_listeners();
diff --git a/replica/database.hh b/replica/database.hh
index a0c5850dc5..338e082bd1 100644
--- a/replica/database.hh
+++ b/replica/database.hh
@@ -416,9 +416,6 @@ public:
         bool enable_node_aggregated_table_metrics = true;
         size_t view_update_concurrency_semaphore_limit;
         db::data_listeners* data_listeners = nullptr;
-        // Not really table-specific (it's a global configuration parameter), but stored here
-        // for easy access from `table` member functions:
-        utils::updateable_value<bool> reversed_reads_auto_bypass_cache{false};
         uint32_t tombstone_warn_threshold{0};
         unsigned x_log2_compaction_groups{0};
         utils::updateable_value<bool> enable_compacting_data_for_streaming_and_repair;
diff --git a/replica/table.cc b/replica/table.cc
index 0bd5cf636a..2b7ec8483a 100644
--- a/replica/table.cc
+++ b/replica/table.cc
@@ -246,7 +246,7 @@ table::make_reader_v2(schema_ptr s,
     });
 
     const auto bypass_cache = slice.options.contains(query::partition_slice::option::bypass_cache);
-    if (cache_enabled() && !bypass_cache && !(slice.is_reversed() && _config.reversed_reads_auto_bypass_cache())) {
+    if (cache_enabled() && !bypass_cache) {
         if (auto reader_opt = _cache.make_reader_opt(s, permit, range, slice, &_compaction_manager.get_tombstone_gc_state(), std::move(trace_state), fwd, fwd_mr)) {
             readers.emplace_back(std::move(*reader_opt));
         }