diff --git a/index-cid-to-subset-offset.go b/index-cid-to-subset-offset.go
index 84512f08..48f93613 100644
--- a/index-cid-to-subset-offset.go
+++ b/index-cid-to-subset-offset.go
@@ -12,8 +12,10 @@ import (
 
 	"github.com/davecgh/go-spew/spew"
 	"github.com/dustin/go-humanize"
+	"github.com/ipfs/go-cid"
 	"github.com/ipld/go-car/util"
 	carv2 "github.com/ipld/go-car/v2"
+	cidlink "github.com/ipld/go-ipld-prime/linking/cid"
 	"github.com/rpcpool/yellowstone-faithful/carreader"
 	"github.com/rpcpool/yellowstone-faithful/indexes"
 	"github.com/rpcpool/yellowstone-faithful/iplddecoders"
@@ -29,6 +31,8 @@ func CreateIndex_cid2subsetOffset(
 	indexDir string,
 ) (string, error) {
 	var numItems uint64
+	var orderedCids []cid.Cid
+	carPathMap := make(map[cid.Cid]string)
 	for _, carPath := range carPaths {
 		// Check if the CAR file exists:
 		exists, err := fileExists(carPath)
@@ -39,11 +43,34 @@ func CreateIndex_cid2subsetOffset(
 			return "", fmt.Errorf("CAR file %q does not exist", carPath)
 		}
 
+		carFile, err := os.Open(carPath)
+		if err != nil {
+			return "", fmt.Errorf("failed to open car file: %w", err)
+		}
+		defer carFile.Close()
+
+		rd, err := carreader.New(carFile)
+		if err != nil {
+			return "", fmt.Errorf("failed to create car reader: %w", err)
+		}
+
+		// check it has 1 root
+		if len(rd.Header.Roots) != 1 {
+			return "", fmt.Errorf("car file must have exactly 1 root, but has %d", len(rd.Header.Roots))
+		}
+
+		carPathMap[rd.Header.Roots[0]] = carPath
+
 		klog.Infof("Counting items in car file: %s", carPath)
-		ni, err := carCountItems(carPath)
+		ni, epochObject, err := carCountItemsAndFindEpoch(carPath)
 		if err != nil {
 			return "", fmt.Errorf("failed to count items in CAR: %w", err)
 		}
+		if epochObject != nil {
+			for _, subset := range epochObject.Subsets {
+				orderedCids = append(orderedCids, subset.(cidlink.Link).Cid)
+			}
+		}
 		klog.Infof("Found %s items in car file", humanize.Comma(int64(ni)))
 		numItems += ni
 	}
@@ -70,7 +97,8 @@ func CreateIndex_cid2subsetOffset(
 	// To do: how to get the subset index?
 	subset := uint64(0)
 	numItemsIndexed := uint64(0)
-	for _, carPath := range carPaths {
+	for _, cid := range orderedCids {
+		carPath := carPathMap[cid]
 		carFile, err := os.Open(carPath)
 		if err != nil {
 			return "", fmt.Errorf("failed to open car file: %w", err)
@@ -94,6 +122,7 @@ func CreateIndex_cid2subsetOffset(
 			c, sectionLength, err := rd.NextInfo()
 			if err != nil {
 				if errors.Is(err, io.EOF) {
+					subset++
 					break
 				}
 				return "", fmt.Errorf("encountered an error while indexing: %w", err)
diff --git a/readers.go b/readers.go
index 2b796a72..591fbe2f 100644
--- a/readers.go
+++ b/readers.go
@@ -62,6 +62,43 @@ func carCountItems(carPath string) (uint64, error) {
 	return count, nil
 }
 
+func carCountItemsAndFindEpoch(carPath string) (uint64, *ipldbindcode.Epoch, error) {
+	file, err := os.Open(carPath)
+	if err != nil {
+		return 0, nil, err
+	}
+	defer file.Close()
+
+	rd, err := carreader.New(file)
+	if err != nil {
+		return 0, nil, fmt.Errorf("failed to open car file: %w", err)
+	}
+
+	var count uint64
+	var epochObject *ipldbindcode.Epoch
+	for {
+		_, _, block, err := rd.NextNodeBytes()
+		if err != nil {
+			if errors.Is(err, io.EOF) {
+				break
+			}
+			return 0, nil, err
+		}
+		count++
+
+		fdb := block[1]
+		if iplddecoders.Kind(fdb) == iplddecoders.KindEpoch {
+			epochObject, err = iplddecoders.DecodeEpoch(block)
+			if err != nil {
+				return 0, nil, fmt.Errorf("failed to decode Epoch node: %w", err)
+			}
+		}
+	}
+
+	return count, epochObject, nil
+
+}
+
 func carCountItemsByFirstByte(carPath string) (map[byte]uint64, *ipldbindcode.Epoch, error) {
 	file, err := os.Open(carPath)
 	if err != nil {