diff --git a/src/HtmlParser.js b/src/HtmlParser.js
index ab16a8f..5d3b730 100644
--- a/src/HtmlParser.js
+++ b/src/HtmlParser.js
@@ -11,6 +11,8 @@ import {escapeQuotes} from './utils';
 
  * @type {Object}
  */
+const ignoredInvalidTags = /^<\s*(?:script|style)/i;
+
 const detect = {
   comment: /^<!--/,
   endTag: /^<\//,
@@ -30,12 +32,14 @@ export default class HtmlParser {
    * @param {string} stream The initial parse stream contents.
    * @param {Object} options The options
    * @param {boolean} options.autoFix Set to true to automatically fix errors
+   * @param {boolean} options.allowInvalidHTML Continue parsing invalid HTML, dropping minimal characters
    */
   constructor(stream = '', options = {}) {
     this.stream = stream;
 
     let fix = false;
     const fixedTokenOptions = {};
+    this.allowInvalidHTML = options.allowInvalidHTML;
 
     for (let key in supports) {
       if (supports.hasOwnProperty(key)) {
@@ -48,10 +52,10 @@ export default class HtmlParser {
 
     if (fix) {
       this._readToken = fixedReadTokenFactory(this, fixedTokenOptions, () => this._readTokenImpl());
-      this._peekToken = fixedReadTokenFactory(this, fixedTokenOptions, () => this._peekTokenImpl());
+      this._peekToken = fixedReadTokenFactory(this, fixedTokenOptions, () => this._peekTokenImpl(this.allowInvalidHTML));
     } else {
       this._readToken = this._readTokenImpl;
-      this._peekToken = this._peekTokenImpl;
+      this._peekToken = () => this._peekTokenImpl(this.allowInvalidHTML);
     }
   }
 
@@ -80,19 +84,31 @@ export default class HtmlParser {
    * @returns {?Token}
    */
   _readTokenImpl() {
-    const token = this._peekTokenImpl();
+    const token = this._peekTokenImpl(this.allowInvalidHTML);
     if (token) {
       this.stream = this.stream.slice(token.length);
       return token;
     }
   }
 
+  _consumeInvalidHtml() {
+    this.stream = this.stream.slice(1);
+
+    let token = this._peekTokenImpl(false);
+    while (this.stream.length > 0 && token == null) {
+      this.stream = this.stream.slice(1);
+      token = this._peekTokenImpl(false);
+    }
+
+    return token;
+  }
+
   /**
    * The implementation of token peeking.
    *
    * @returns {?Token}
    */
-  _peekTokenImpl() {
+  _peekTokenImpl(allowInvalidHTML) {
     for (let type in detect) {
       if (detect.hasOwnProperty(type)) {
         if (detect[type].test(this.stream)) {
@@ -106,6 +122,8 @@ export default class HtmlParser {
               token.text = this.stream.substr(0, token.length);
               return token;
             }
+          } else if (allowInvalidHTML && this.stream.length && token == null && !(ignoredInvalidTags.test(this.stream))) {
+            return this._consumeInvalidHtml();
           }
         }
       }
diff --git a/src/fixedReadTokenFactory.js b/src/fixedReadTokenFactory.js
index f80ed93..01edfae 100644
--- a/src/fixedReadTokenFactory.js
+++ b/src/fixedReadTokenFactory.js
@@ -10,7 +10,7 @@ const EMPTY = /^(AREA|BASE|BASEFONT|BR|COL|FRAME|HR|IMG|INPUT|ISINDEX|LINK|META|
  *
  * @type {RegExp}
  */
-const CLOSESELF = /^(COLGROUP|DD|DT|LI|OPTIONS|P|TD|TFOOT|TH|THEAD|TR)$/i;
+const CLOSESELF = /^(COLGROUP|DD|DT|LI|OPTIONS|P|TD|TFOOT|TH|THEAD|TR|BR)$/i;
 
 /**
  * Corrects a token.
diff --git a/test/helpers/testParse.js b/test/helpers/testParse.js
index c5f06c2..ce8e7cc 100644
--- a/test/helpers/testParse.js
+++ b/test/helpers/testParse.js
@@ -54,7 +54,7 @@ export function parsesCompletely(s, options, theTest) {
 
 export function fixes(s, theTest) {
   return () => {
-    const parser = typeof s === 'string' ? new HtmlParser(s, {autoFix: true}) : s;
+    const parser = typeof s === 'string' ? new HtmlParser(s, {autoFix: true, allowInvalidHTML: true}) : s;
     let tok = parser.readToken();
     let str = '';
     while (tok) {
diff --git a/test/unit/tags.spec.js b/test/unit/tags.spec.js
index fee4449..76e4511 100644
--- a/test/unit/tags.spec.js
+++ b/test/unit/tags.spec.js
@@ -118,6 +118,10 @@ describe('HtmlParser (tags)', () => {
       expect(str).to.equal('</DIV>');
     });
 
+    it('recovers from broken html', fixes('<div> hello <br \\="" /> </div><h2>Example</h2>', s => {
+      expect(s).to.equal('<div> hello br \\="" /> </div><h2>Example</h2>');
+    }));
+
     it('fixes missing end tag', fixes('<div><i></div>foo', s => {
       expect(s).to.equal('<div><i></i></div>foo');
     }));