diff --git a/txtar/archive.go b/txtar/archive.go
index fd95f1e64a1..72d0b4fe7f3 100644
--- a/txtar/archive.go
+++ b/txtar/archive.go
@@ -53,17 +53,42 @@ type File struct {
 // Format returns the serialized form of an Archive.
 // It is assumed that the Archive data structure is well-formed:
 // a.Comment and all a.File[i].Data contain no file marker lines,
-// and all a.File[i].Name is non-empty.
+// and all a.File[i].Name is non-empty. Format uses line separators
+// based on the line separator encountered in the comment section.
 func Format(a *Archive) []byte {
+	firstSep, ok := lineEnd(a.Comment)
+	if !ok {
+		for _, f := range a.Files {
+			firstSep, ok = lineEnd(f.Data)
+			if ok {
+				break
+			}
+		}
+	}
+
 	var buf bytes.Buffer
-	buf.Write(fixNL(a.Comment))
+	buf.Write(fixNL(a.Comment, firstSep))
 	for _, f := range a.Files {
-		fmt.Fprintf(&buf, "-- %s --\n", f.Name)
-		buf.Write(fixNL(f.Data))
+		fmt.Fprintf(&buf, "-- %s --%s", f.Name, firstSep)
+		buf.Write(fixNL(f.Data, firstSep))
 	}
 	return buf.Bytes()
 }
 
+// lineEnd returns the line separator that was used at the end
+// of the passed byte slice and a bool flag that says if the line separator
+// was really found or it is a default value (lf).
+func lineEnd(b []byte) ([]byte, bool) {
+	switch {
+	case bytes.HasSuffix(b, crlf):
+		return crlf, true
+	case bytes.HasSuffix(b, lf):
+		return lf, true
+	default:
+		return lf, false
+	}
+}
+
 // ParseFile parses the named file as an archive.
 func ParseFile(file string) (*Archive, error) {
 	data, err := os.ReadFile(file)
@@ -78,16 +103,19 @@ func ParseFile(file string) (*Archive, error) {
 func Parse(data []byte) *Archive {
 	a := new(Archive)
 	var name string
-	a.Comment, name, data = findFileMarker(data)
+	var lineSeparator []byte
+	a.Comment, name, lineSeparator, data = findFileMarker(data, nil)
 	for name != "" {
 		f := File{name, nil}
-		f.Data, name, data = findFileMarker(data)
+		f.Data, name, lineSeparator, data = findFileMarker(data, lineSeparator)
 		a.Files = append(a.Files, f)
 	}
 	return a
 }
 
 var (
+	crlf          = []byte("\r\n")
+	lf            = []byte("\n")
 	newlineMarker = []byte("\n-- ")
 	marker        = []byte("-- ")
 	markerEnd     = []byte(" --")
@@ -96,45 +124,55 @@ var (
 // findFileMarker finds the next file marker in data,
 // extracts the file name, and returns the data before the marker,
 // the file name, and the data after the marker.
+// lineSep states if \n or \r\n should be appended by fixNL.
 // If there is no next marker, findFileMarker returns before = fixNL(data), name = "", after = nil.
-func findFileMarker(data []byte) (before []byte, name string, after []byte) {
+func findFileMarker(data, lineSep []byte) (before []byte, name string, lineSeparator []byte, after []byte) {
 	var i int
 	for {
-		if name, after = isMarker(data[i:]); name != "" {
-			return data[:i], name, after
+		if name, lineSeparator, after = isMarker(data[i:]); name != "" {
+			return data[:i], name, lineSeparator, after
 		}
 		j := bytes.Index(data[i:], newlineMarker)
 		if j < 0 {
-			return fixNL(data), "", nil
+			return fixNL(data, lineSep), "", lineSep, nil
 		}
 		i += j + 1 // positioned at start of new possible marker
 	}
 }
 
 // isMarker checks whether data begins with a file marker line.
-// If so, it returns the name from the line and the data after the line.
-// Otherwise it returns name == "" with an unspecified after.
-func isMarker(data []byte) (name string, after []byte) {
+// If so, it returns the name from the line, used line separator and the data after the line.
+// Otherwise it returns name == "" with nil lineSeparator ("\n") and after.
+// If the data does not contain a new line the default line separator is returned ("\n").
+func isMarker(data []byte) (name string, lineSeparator, after []byte) {
 	if !bytes.HasPrefix(data, marker) {
-		return "", nil
+		return "", nil, nil
 	}
+	lineSeparator = lf
 	if i := bytes.IndexByte(data, '\n'); i >= 0 {
-		data, after = data[:i], data[i+1:]
+		if len(data) > 0 && data[i-1] == '\r' {
+			data, after = data[:i-1], data[i+1:]
+			lineSeparator = crlf
+		} else {
+			data, after = data[:i], data[i+1:]
+			lineSeparator = lf
+		}
 	}
 	if !(bytes.HasSuffix(data, markerEnd) && len(data) >= len(marker)+len(markerEnd)) {
-		return "", nil
+		return "", nil, nil
 	}
-	return strings.TrimSpace(string(data[len(marker) : len(data)-len(markerEnd)])), after
+	return strings.TrimSpace(string(data[len(marker) : len(data)-len(markerEnd)])), lineSeparator, after
 }
 
-// If data is empty or ends in \n, fixNL returns data.
-// Otherwise fixNL returns a new slice consisting of data with a final \n added.
-func fixNL(data []byte) []byte {
-	if len(data) == 0 || data[len(data)-1] == '\n' {
+// If data is empty or ends in lineSeparator, fixNL returns data.
+// lineSeparator states if \n or \r\n should be appended as a line separator if it is not present.
+// Otherwise fixNL returns a new slice consisting of data with a final lineSeparator added.
+func fixNL(data , lineSeparator []byte) []byte {
+	if len(data) == 0 || bytes.HasSuffix(data, lf) {
 		return data
 	}
-	d := make([]byte, len(data)+1)
+	d := make([]byte, len(data)+len(lineSeparator))
 	copy(d, data)
-	d[len(data)] = '\n'
+	copy(d[len(data):], lineSeparator)
 	return d
 }
diff --git a/txtar/archive_test.go b/txtar/archive_test.go
index 6534f530103..009d0d63c3a 100644
--- a/txtar/archive_test.go
+++ b/txtar/archive_test.go
@@ -18,7 +18,7 @@ func TestParse(t *testing.T) {
 		parsed *Archive
 	}{
 		{
-			name: "basic",
+			name: "basic with line ending LF",
 			text: `comment1
 comment2
 -- file1 --
@@ -33,14 +33,73 @@ hello world
 -- empty filename line --
 some content
 -- --`,
+parsed: &Archive{
+	Comment: []byte("comment1\ncomment2\n"),
+	Files: []File{
+		{"file1", []byte("File 1 text.\n-- foo ---\nMore file 1 text.\n")},
+		{"file 2", []byte("File 2 text.\n")},
+		{"empty", []byte{}},
+		{"noNL", []byte("hello world\n")},
+		{"empty filename line", []byte("some content\n-- --\n")},
+	},
+},
+		},
+		{
+			name: "basic with line ending CRLF",
+			text: "comment1\r\n" +
+			"comment2\r\n" +
+			"-- file1 --\r\n" +
+			"File 1 text.\r\n" +
+			"-- foo ---\r\n" +
+			"More file 1 text.\r\n" +
+			"-- file 2 --\r\n" +
+			"File 2 text.\r\n" +
+			"-- empty --\r\n" +
+			"-- noNL --\r\n" +
+			"hello world\r\n" +
+			"-- empty filename line --\r\n" +
+			"some content\r\n" +
+			"-- --\r\n",
 			parsed: &Archive{
-				Comment: []byte("comment1\ncomment2\n"),
+				Comment: []byte("comment1\r\ncomment2\r\n"),
 				Files: []File{
-					{"file1", []byte("File 1 text.\n-- foo ---\nMore file 1 text.\n")},
-					{"file 2", []byte("File 2 text.\n")},
+					{"file1", []byte("File 1 text.\r\n-- foo ---\r\nMore file 1 text.\r\n")},
+					{"file 2", []byte("File 2 text.\r\n")},
+					{"empty", []byte{}},
+					{"noNL", []byte("hello world\r\n")},
+					{"empty filename line", []byte("some content\r\n-- --\r\n")},
+				},
+			},
+		},
+		{
+			name: "mixed line endings",
+			text: "comment1\n" +
+			"comment2\r\n" +
+			"-- file1 --\r\n" +
+			"File 1 text.\n" +
+			"-- foo ---\r\n" +
+			"More file 1 text.\r\n" +
+			"-- file 2 --\r\n" +
+			"File 2 text.\r\n" +
+			"-- file 3 --\r\n" +
+			"File 3 text.\r\n" +
+			"-- foo ---\r\n" +
+			"More file 3 text.\r\n" +
+			"-- empty --\r\n" +
+			"-- noNL --\r\n" +
+			"hello world\r\n" +
+			"-- empty filename line --\r\n" +
+			"some content\r\n" +
+			"-- --\n",
+			parsed: &Archive{
+				Comment: []byte("comment1\ncomment2\r\n"),
+				Files: []File{
+					{"file1", []byte("File 1 text.\n-- foo ---\r\nMore file 1 text.\r\n")},
+					{"file 2", []byte("File 2 text.\r\n")},
+					{"file 3", []byte("File 3 text.\r\n-- foo ---\r\nMore file 3 text.\r\n")},
 					{"empty", []byte{}},
-					{"noNL", []byte("hello world\n")},
-					{"empty filename line", []byte("some content\n-- --\n")},
+					{"noNL", []byte("hello world\r\n")},
+					{"empty filename line", []byte("some content\r\n-- --\n")},
 				},
 			},
 		},
@@ -67,7 +126,7 @@ func TestFormat(t *testing.T) {
 		wanted string
 	}{
 		{
-			name: "basic",
+			name: "basic with line ending LF",
 			input: &Archive{
 				Comment: []byte("comment1\ncomment2\n"),
 				Files: []File{
@@ -90,6 +149,61 @@ File 2 text.
 hello world
 `,
 		},
+		{
+			name: "basic with line ending CRLF",
+			input: &Archive{
+				Comment: []byte("comment1\r\ncomment2\r\n"),
+				Files: []File{
+					{"file1", []byte("File 1 text.\r\n-- foo ---\r\nMore file 1 text.\r\n")},
+					{"file 2", []byte("File 2 text.\r\n")},
+					{"empty", []byte{}},
+					{"noNL", []byte("hello world")},
+				},
+			},
+			wanted: "comment1\r\n" +
+				"comment2\r\n" +
+				"-- file1 --\r\n" +
+				"File 1 text.\r\n" +
+				"-- foo ---\r\n" +
+				"More file 1 text.\r\n" +
+				"-- file 2 --\r\n" +
+				"File 2 text.\r\n" +
+				"-- empty --\r\n" +
+				"-- noNL --\r\n" +
+				"hello world\r\n",
+		},
+		{
+			name: "mixed line endings",
+			input: &Archive{
+				Comment: []byte("comment1\ncomment2\r\n"),
+				Files: []File{
+					{"file1", []byte("File 1 text.\n-- foo ---\r\nMore file 1 text.\r\n")},
+					{"file 2", []byte("File 2 text.\r\n")},
+					{"file 3", []byte("File 3 text.\r\n-- foo ---\r\nMore file 3 text.\r\n")},
+					{"empty", []byte{}},
+					{"noNL", []byte("hello world\r\n")},
+					{"empty filename line", []byte("some content\r\n-- --\n")},
+				},
+			},
+			wanted: "comment1\n" +
+				"comment2\r\n" +
+				"-- file1 --\r\n" +
+				"File 1 text.\n" +
+				"-- foo ---\r\n" +
+				"More file 1 text.\r\n" +
+				"-- file 2 --\r\n" +
+				"File 2 text.\r\n" +
+				"-- file 3 --\r\n" +
+				"File 3 text.\r\n" +
+				"-- foo ---\r\n" +
+				"More file 3 text.\r\n" +
+				"-- empty --\r\n" +
+				"-- noNL --\r\n" +
+				"hello world\r\n" +
+				"-- empty filename line --\r\n" +
+				"some content\r\n" +
+				"-- --\n",
+		},
 	}
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {