diff --git a/lib/dash/dash_parser.js b/lib/dash/dash_parser.js
index 431419160..b298d828c 100644
--- a/lib/dash/dash_parser.js
+++ b/lib/dash/dash_parser.js
@@ -423,7 +423,7 @@ shaka.dash.DashParser.prototype.parseManifest_ =
   var Functional = shaka.util.Functional;
   var XmlUtils = shaka.util.XmlUtils;
 
-  var string = shaka.util.StringUtils.fromBytesAutoDetect(data);
+  var string = shaka.util.StringUtils.fromUTF8(data);
   var parser = new DOMParser();
   var xml = null;
   var mpd = null;
@@ -1174,7 +1174,7 @@ shaka.dash.DashParser.prototype.requestForTiming_ =
 
       text = response.headers['date'];
     } else {
-      text = shaka.util.StringUtils.fromBytesAutoDetect(response.data);
+      text = shaka.util.StringUtils.fromUTF8(response.data);
     }
 
     var date = Date.parse(text);
diff --git a/lib/media/mp4_vtt_parser.js b/lib/media/mp4_vtt_parser.js
index 11155edb7..6f70bc6b1 100644
--- a/lib/media/mp4_vtt_parser.js
+++ b/lib/media/mp4_vtt_parser.js
@@ -121,7 +121,7 @@ shaka.media.Mp4VttParser.parseCue_ = function(
     var startPosition = reader.getPosition();
     var size = reader.readUint32();
     var type = reader.readUint32();
-    var content = shaka.util.StringUtils.fromBytesAutoDetect(
+    var content = shaka.util.StringUtils.fromUTF8(
         reader.readBytes(size - 8).buffer);
     if (size == 1) {
       size = reader.readUint64();
diff --git a/lib/media/ttml_text_parser.js b/lib/media/ttml_text_parser.js
index c9a261ea9..3dd62cfc2 100644
--- a/lib/media/ttml_text_parser.js
+++ b/lib/media/ttml_text_parser.js
@@ -33,7 +33,7 @@ goog.require('shaka.util.StringUtils');
  * @throws {shaka.util.Error}
  */
 shaka.media.TtmlTextParser = function(data, segmentStartTime, segmentEndTime) {
-  var str = shaka.util.StringUtils.fromBytesAutoDetect(data);
+  var str = shaka.util.StringUtils.fromUTF8(data);
   var ret = [];
   var parser = new DOMParser();
   var xml = null;
diff --git a/lib/media/vtt_text_parser.js b/lib/media/vtt_text_parser.js
index 197c8f8b7..f03a4f607 100644
--- a/lib/media/vtt_text_parser.js
+++ b/lib/media/vtt_text_parser.js
@@ -34,7 +34,7 @@ goog.require('shaka.util.TextParser');
  */
 shaka.media.VttTextParser = function(data, segmentStartTime, segmentEndTime) {
   // Get the input as a string.  Normalize newlines to \n.
-  var str = shaka.util.StringUtils.fromBytesAutoDetect(data);
+  var str = shaka.util.StringUtils.fromUTF8(data);
   str = str.replace(/\r\n|\r(?=[^\n]|$)/gm, '\n');
   var blocks = str.split(/\n{2,}/m);
 
diff --git a/lib/util/data_view_reader.js b/lib/util/data_view_reader.js
index 1ff68408f..8b0b4b2d6 100644
--- a/lib/util/data_view_reader.js
+++ b/lib/util/data_view_reader.js
@@ -197,7 +197,8 @@ shaka.util.DataViewReader.prototype.skip = function(bytes) {
 
 
 /**
- * Keeps reading until it reaches a byte that equals to zero.
+ * Keeps reading until it reaches a byte that equals to zero.  The text is
+ * assumed to be UTF-8.
  * @return {string}
  * @throws {shaka.util.Error} when reading past the end of the data view.
  */
@@ -216,7 +217,7 @@ shaka.util.DataViewReader.prototype.readTerminatedString = function() {
   var ret = this.dataView_.buffer.slice(start, this.position_);
   // skip string termination
   this.position_ += 1;
-  return shaka.util.StringUtils.fromBytesAutoDetect(ret);
+  return shaka.util.StringUtils.fromUTF8(ret);
 };
 
 
diff --git a/lib/util/string_utils.js b/lib/util/string_utils.js
index 64f09bee6..5c1d26e79 100644
--- a/lib/util/string_utils.js
+++ b/lib/util/string_utils.js
@@ -36,8 +36,15 @@ goog.require('shaka.util.Error');
  */
 shaka.util.StringUtils.fromUTF8 = function(data) {
   if (!data) return '';
+
+  var uint8 = new Uint8Array(data);
+  // If present, strip off the UTF-8 BOM.
+  if (uint8[0] == 0xef && uint8[1] == 0xbb && uint8[2] == 0xbf) {
+    uint8 = uint8.subarray(3);
+  }
+
   // http://stackoverflow.com/a/13691499
-  var utf8 = shaka.util.StringUtils.fromCharCode_(new Uint8Array(data));
+  var utf8 = shaka.util.StringUtils.fromCharCode_(uint8);
   // This converts each character in the string to an escape sequence.  If the
   // character is in the ASCII range, it is not converted; otherwise it is
   // converted to a URI escape sequence.
@@ -109,7 +116,7 @@ shaka.util.StringUtils.fromBytesAutoDetect = function(data) {
 
   var uint8 = new Uint8Array(data);
   if (uint8[0] == 0xef && uint8[1] == 0xbb && uint8[2] == 0xbf)
-    return StringUtils.fromUTF8(uint8.subarray(3));
+    return StringUtils.fromUTF8(uint8);
   else if (uint8[0] == 0xfe && uint8[1] == 0xff)
     return StringUtils.fromUTF16(uint8.subarray(2), false /* littleEndian */);
   else if (uint8[0] == 0xff && uint8[1] == 0xfe)
diff --git a/test/util/string_utils_unit.js b/test/util/string_utils_unit.js
index 3ddc6e87c..2834a6fb0 100644
--- a/test/util/string_utils_unit.js
+++ b/test/util/string_utils_unit.js
@@ -30,6 +30,14 @@ describe('StringUtils', function() {
     expect(StringUtils.fromUTF8(buffer)).toBe('F\u20ac \ud800\udf48');
   });
 
+  it('strips the BOM in fromUTF8', function() {
+    // This is 4 Unicode characters, the last will be split into a surrogate
+    // pair.
+    var arr = [0xef, 0xbb, 0xbf, 0x74, 0x65, 0x78, 0x74];
+    var buffer = new Uint8Array(arr).buffer;
+    expect(StringUtils.fromUTF8(buffer)).toBe('text');
+  });
+
   it('parses fromUTF16 big-endian', function() {
     // This is big-endian pairs of 16-bit numbers.  This translates into 3
     // Unicode characters where the last is split into a surrogate pair.