diff --git a/agent/http.go b/agent/http.go
index f6a8b448e4..ad2bbe9779 100644
--- a/agent/http.go
+++ b/agent/http.go
@@ -723,6 +723,13 @@ func setMeta(resp http.ResponseWriter, m structs.QueryMetaCompat) {
 	setLastContact(resp, m.GetLastContact())
 	setKnownLeader(resp, m.GetKnownLeader())
 	setConsistency(resp, m.GetConsistencyLevel())
+	setQueryBackend(resp, m.GetBackend())
+}
+
+func setQueryBackend(resp http.ResponseWriter, backend structs.QueryBackend) {
+	if b := backend.String(); b != "" {
+		resp.Header().Set("X-Consul-Query-Backend", b)
+	}
 }
 
 // setCacheMeta sets http response headers to indicate cache status.
diff --git a/agent/structs/protobuf_compat.go b/agent/structs/protobuf_compat.go
index 5322288264..667443c9e9 100644
--- a/agent/structs/protobuf_compat.go
+++ b/agent/structs/protobuf_compat.go
@@ -44,6 +44,7 @@ type QueryMetaCompat interface {
 	SetIndex(uint64)
 	GetConsistencyLevel() string
 	SetConsistencyLevel(string)
+	GetBackend() QueryBackend
 }
 
 // GetToken helps implement the QueryOptionsCompat interface
@@ -269,3 +270,7 @@ func (q *QueryMeta) SetIndex(index uint64) {
 func (q *QueryMeta) SetConsistencyLevel(consistencyLevel string) {
 	q.ConsistencyLevel = consistencyLevel
 }
+
+func (q *QueryMeta) GetBackend() QueryBackend {
+	return q.Backend
+}
diff --git a/agent/structs/structs.go b/agent/structs/structs.go
index 93c816e4b6..a0d3e10338 100644
--- a/agent/structs/structs.go
+++ b/agent/structs/structs.go
@@ -339,6 +339,24 @@ func (w WriteRequest) HasTimedOut(start time.Time, rpcHoldTimeout, maxQueryTime,
 	return time.Since(start) > rpcHoldTimeout
 }
 
+type QueryBackend int
+
+const (
+	QueryBackendBlocking QueryBackend = iota
+	QueryBackendStreaming
+)
+
+func (q QueryBackend) String() string {
+	switch q {
+	case QueryBackendBlocking:
+		return "blocking-query"
+	case QueryBackendStreaming:
+		return "streaming"
+	default:
+		return ""
+	}
+}
+
 // QueryMeta allows a query response to include potentially
 // useful metadata about a query
 type QueryMeta struct {
@@ -363,6 +381,9 @@ type QueryMeta struct {
 	// When NotModified is true, the response will not contain the result of
 	// the query.
 	NotModified bool
+
+	// Backend used to handle this query, either blocking-query or streaming.
+	Backend QueryBackend
 }
 
 // RegisterRequest is used for the Catalog.Register endpoint
diff --git a/proto/pbcommon/common.go b/proto/pbcommon/common.go
index e6e981ccfa..eb396ae58e 100644
--- a/proto/pbcommon/common.go
+++ b/proto/pbcommon/common.go
@@ -2,6 +2,8 @@ package pbcommon
 
 import (
 	"time"
+
+	"github.com/hashicorp/consul/agent/structs"
 )
 
 // IsRead is always true for QueryOption
@@ -97,6 +99,10 @@ func (q *QueryMeta) SetConsistencyLevel(consistencyLevel string) {
 	q.ConsistencyLevel = consistencyLevel
 }
 
+func (q *QueryMeta) GetBackend() structs.QueryBackend {
+	return structs.QueryBackend(0)
+}
+
 // WriteRequest only applies to writes, always false
 func (w WriteRequest) IsRead() bool {
 	return false
diff --git a/website/content/api-docs/features/blocking.mdx b/website/content/api-docs/features/blocking.mdx
index 2b7a9ae860..4087c64d55 100644
--- a/website/content/api-docs/features/blocking.mdx
+++ b/website/content/api-docs/features/blocking.mdx
@@ -99,6 +99,9 @@ While streaming is a significant optimization over long polling, it will not pop
 `X-Consul-LastContact` or `X-Consul-KnownLeader` response headers, because the required
 data is not available to the client.
 
+When the streaming backend is used, API responses will include the `X-Consul-Query-Backend`
+header with a value of `streaming`.
+
 
 ## Hash-based Blocking Queries