diff --git a/src/query/api/v1/handler/prometheus/handleroptions/fetch_options.go b/src/query/api/v1/handler/prometheus/handleroptions/fetch_options.go
index a812d19828..d30e631be8 100644
--- a/src/query/api/v1/handler/prometheus/handleroptions/fetch_options.go
+++ b/src/query/api/v1/handler/prometheus/handleroptions/fetch_options.go
@@ -134,6 +134,20 @@ func ParseLimit(req *http.Request, header, formValue string, defaultLimit int) (
 	return defaultLimit, nil
 }
 
+// ParseInstanceMultiple parses request instance multiple from header.
+func ParseInstanceMultiple(req *http.Request, defaultValue float32) (float32, error) {
+	if str := req.Header.Get(headers.LimitInstanceMultipleHeader); str != "" {
+		v, err := strconv.ParseFloat(str, 32)
+		if err != nil {
+			err = fmt.Errorf(
+				"could not parse instance multiple: input=%s, err=%w", str, err)
+			return 0, err
+		}
+		return float32(v), nil
+	}
+	return defaultValue, nil
+}
+
 // ParseRequireExhaustive parses request limit require exhaustive from header or
 // query string.
 func ParseRequireExhaustive(req *http.Request, defaultValue bool) (bool, error) {
@@ -214,9 +228,13 @@ func (b fetchOptionsBuilder) newFetchOptions(
 	if err != nil {
 		return nil, nil, err
 	}
-
 	fetchOpts.SeriesLimit = seriesLimit
-	fetchOpts.InstanceMultiple = b.opts.Limits.InstanceMultiple
+
+	instanceMultiple, err := ParseInstanceMultiple(req, b.opts.Limits.InstanceMultiple)
+	if err != nil {
+		return nil, nil, err
+	}
+	fetchOpts.InstanceMultiple = instanceMultiple
 
 	docsLimit, err := ParseLimit(req, headers.LimitMaxDocsHeader,
 		"docsLimit", b.opts.Limits.DocsLimit)
diff --git a/src/query/api/v1/handler/prometheus/handleroptions/fetch_options_test.go b/src/query/api/v1/handler/prometheus/handleroptions/fetch_options_test.go
index f2b4dfae2c..312d8589f3 100644
--- a/src/query/api/v1/handler/prometheus/handleroptions/fetch_options_test.go
+++ b/src/query/api/v1/handler/prometheus/handleroptions/fetch_options_test.go
@@ -475,3 +475,20 @@ func TestTimeoutParseWithGetRequestParam(t *testing.T) {
 	assert.NoError(t, err)
 	assert.Equal(t, timeout, time.Millisecond)
 }
+
+func TestInstanceMultiple(t *testing.T) {
+	req := httptest.NewRequest("GET", "/", nil)
+	m, err := ParseInstanceMultiple(req, 2.0)
+	require.NoError(t, err)
+	require.Equal(t, float32(2.0), m)
+
+	req.Header.Set(headers.LimitInstanceMultipleHeader, "3.0")
+	m, err = ParseInstanceMultiple(req, 2.0)
+	require.NoError(t, err)
+	require.Equal(t, float32(3.0), m)
+
+	req.Header.Set(headers.LimitInstanceMultipleHeader, "blah")
+	_, err = ParseInstanceMultiple(req, 2.0)
+	require.Error(t, err)
+	require.Contains(t, err.Error(), "could not parse instance multiple")
+}
diff --git a/src/x/headers/headers.go b/src/x/headers/headers.go
index 59cede447d..76368b7320 100644
--- a/src/x/headers/headers.go
+++ b/src/x/headers/headers.go
@@ -84,6 +84,9 @@ const (
 	// the number of time series returned by each storage node.
 	LimitMaxSeriesHeader = M3HeaderPrefix + "Limit-Max-Series"
 
+	// LimitInstanceMultipleHeader overrides the PerQueryLimitsConfiguration.InstanceMultiple for the request.
+	LimitInstanceMultipleHeader = M3HeaderPrefix + "Limit-Instance-Multiple"
+
 	// LimitMaxDocsHeader is the M3 limit docs header that limits
 	// the number of docs returned by each storage node.
 	LimitMaxDocsHeader = M3HeaderPrefix + "Limit-Max-Docs"