diff --git a/ext/ragel/base_lexer.rl b/ext/ragel/base_lexer.rl
index 3f59b46..ab043d4 100644
--- a/ext/ragel/base_lexer.rl
+++ b/ext/ragel/base_lexer.rl
@@ -98,6 +98,40 @@
         any;
     *|;
 
+    # Processing Instructions
+    #
+    # http://www.w3.org/TR/xpath/#section-Processing-Instruction-Nodes
+    # http://en.wikipedia.org/wiki/Processing_Instruction
+    #
+    # These are tags meant to be used by parsers/libraries for custom behaviour.
+    # One example are the tags used by PHP: <?php and ?>. Note that the XML
+    # declaration tags (<?xml ?>) are not considered to be a processing
+    # instruction.
+    #
+
+    proc_ins_start = '<?' identifier;
+    proc_ins_end   = '?>';
+
+    action start_proc_ins {
+        callback_simple("on_proc_ins_start");
+        callback("on_proc_ins_name", data, encoding, ts + 2, te);
+
+        mark = te;
+
+        fnext proc_ins_body;
+    }
+
+    proc_ins_body := |*
+        proc_ins_end => {
+            callback("on_text", data, encoding, mark, ts);
+            callback_simple("on_proc_ins_end");
+
+            fnext main;
+        };
+
+        any;
+    *|;
+
     # Strings
     #
     # Strings in HTML can either be single or double quoted. If a string
@@ -257,6 +291,7 @@
         xml_decl_start => start_xml_decl;
         comment_start  => start_comment;
         cdata_start    => start_cdata;
+        proc_ins_start => start_proc_ins;
 
         # The start of an element.
         '<' => start_element;
diff --git a/lib/oga/xml/lexer.rb b/lib/oga/xml/lexer.rb
index 485fa17..655d36e 100644
--- a/lib/oga/xml/lexer.rb
+++ b/lib/oga/xml/lexer.rb
@@ -300,6 +300,29 @@ module Oga
         add_token(:T_ELEM_START)
       end
 
+      ##
+      # Called on the start of a processing instruction.
+      #
+      def on_proc_ins_start
+        add_token(:T_PROC_INS_START)
+      end
+
+      ##
+      # Called on a processing instruction name.
+      #
+      # @param [String] value
+      #
+      def on_proc_ins_name(value)
+        add_token(:T_PROC_INS_NAME, value)
+      end
+
+      ##
+      # Called on the end of a processing instruction.
+      #
+      def on_proc_ins_end
+        add_token(:T_PROC_INS_END)
+      end
+
       ##
       # Called on the name of an element.
       #
diff --git a/spec/oga/xml/lexer/processing_instruction_spec.rb b/spec/oga/xml/lexer/processing_instruction_spec.rb
new file mode 100644
index 0000000..80648e8
--- /dev/null
+++ b/spec/oga/xml/lexer/processing_instruction_spec.rb
@@ -0,0 +1,22 @@
+require 'spec_helper'
+
+describe Oga::XML::Lexer do
+  context 'processing instructions' do
+    example 'lex a processing instruction' do
+      lex('<?foo?>').should == [
+        [:T_PROC_INS_START, nil, 1],
+        [:T_PROC_INS_NAME, 'foo', 1],
+        [:T_PROC_INS_END, nil, 1]
+      ]
+    end
+
+    example 'lex a processing instruction containing text' do
+      lex('<?foo bar ?>').should == [
+        [:T_PROC_INS_START, nil, 1],
+        [:T_PROC_INS_NAME, 'foo', 1],
+        [:T_TEXT, ' bar ', 1],
+        [:T_PROC_INS_END, nil, 1]
+      ]
+    end
+  end
+end