diff --git a/pkg/store/bucket.go b/pkg/store/bucket.go
index c2648ed400..147d12609c 100644
--- a/pkg/store/bucket.go
+++ b/pkg/store/bucket.go
@@ -2282,15 +2282,7 @@ func checkNilPosting(l labels.Label, p index.Postings) index.Postings {
 func toPostingGroup(ctx context.Context, lvalsFn func(name string) ([]string, error), m *labels.Matcher) (*postingGroup, error) {
 	if m.Type == labels.MatchRegexp {
 		if vals := findSetMatches(m.Value); len(vals) > 0 {
-			// Sorting will improve the performance dramatically if the dataset is relatively large
-			// since entries in the postings offset table was sorted by label name and value,
-			// the sequential reading is much faster.
-			sort.Strings(vals)
-			toAdd := make([]labels.Label, 0, len(vals))
-			for _, val := range vals {
-				toAdd = append(toAdd, labels.Label{Name: m.Name, Value: val})
-			}
-			return newPostingGroup(false, toAdd, nil), nil
+			return newPostingGroup(false, labelsFromSetMatchers(m.Name, vals), nil), nil
 		}
 	}
 
@@ -2298,12 +2290,29 @@ func toPostingGroup(ctx context.Context, lvalsFn func(name string) ([]string, er
 	// have the label name set too. See: https://github.com/prometheus/prometheus/issues/3575
 	// and https://github.com/prometheus/prometheus/pull/3578#issuecomment-351653555.
 	if m.Matches("") {
+		var toRemove []labels.Label
+
+		// Fast-path for MatchNotRegexp matching.
+		// Inverse of a MatchNotRegexp is MatchRegexp (double negation).
+		// Fast-path for set matching.
+		if m.Type == labels.MatchNotRegexp {
+			if vals := findSetMatches(m.Value); len(vals) > 0 {
+				toRemove = labelsFromSetMatchers(m.Name, vals)
+				return newPostingGroup(true, nil, toRemove), nil
+			}
+		}
+
+		// Fast-path for MatchNotEqual matching.
+		// Inverse of a MatchNotEqual is MatchEqual (double negation).
+		if m.Type == labels.MatchNotEqual {
+			return newPostingGroup(true, nil, []labels.Label{{Name: m.Name, Value: m.Value}}), nil
+		}
+
 		vals, err := lvalsFn(m.Name)
 		if err != nil {
 			return nil, err
 		}
 
-		var toRemove []labels.Label
 		for _, val := range vals {
 			if ctx.Err() != nil {
 				return nil, ctx.Err()
@@ -2339,6 +2348,18 @@ func toPostingGroup(ctx context.Context, lvalsFn func(name string) ([]string, er
 	return newPostingGroup(false, toAdd, nil), nil
 }
 
+func labelsFromSetMatchers(name string, vals []string) []labels.Label {
+	// Sorting will improve the performance dramatically if the dataset is relatively large
+	// since entries in the postings offset table was sorted by label name and value,
+	// the sequential reading is much faster.
+	sort.Strings(vals)
+	toAdd := make([]labels.Label, 0, len(vals))
+	for _, val := range vals {
+		toAdd = append(toAdd, labels.Label{Name: name, Value: val})
+	}
+	return toAdd
+}
+
 type postingPtr struct {
 	keyID int
 	ptr   index.Range
diff --git a/pkg/store/bucket_test.go b/pkg/store/bucket_test.go
index 2f1eb8ef93..186f7499ca 100644
--- a/pkg/store/bucket_test.go
+++ b/pkg/store/bucket_test.go
@@ -1387,6 +1387,8 @@ func benchBucketSeries(t testutil.TB, sampleType chunkenc.ValueType, skipChunk b
 		matchersCase := []*labels.Matcher{
 			labels.MustNewMatcher(labels.MatchEqual, "foo", "bar"),
 			labels.MustNewMatcher(labels.MatchNotEqual, "foo", "bar"),
+			labels.MustNewMatcher(labels.MatchEqual, "j", "0"),
+			labels.MustNewMatcher(labels.MatchNotEqual, "j", "0"),
 			labels.MustNewMatcher(labels.MatchRegexp, "j", "(0|1)"),
 			labels.MustNewMatcher(labels.MatchRegexp, "j", "0|1"),
 			labels.MustNewMatcher(labels.MatchNotRegexp, "j", "(0|1)"),