diff --git a/doc/lgf.dox b/doc/lgf.dox
new file mode 100644
--- /dev/null
+++ b/doc/lgf.dox
@@ -0,0 +1,96 @@
+/* -*- C++ -*-
+ *
+ * This file is a part of LEMON, a generic C++ optimization library
+ *
+ * Copyright (C) 2003-2008
+ * Egervary Jeno Kombinatorikus Optimalizalasi Kutatocsoport
+ * (Egervary Research Group on Combinatorial Optimization, EGRES).
+ *
+ * Permission to use, modify and distribute this software is granted
+ * provided that this copyright notice appears in all copies. For
+ * precise terms see the accompanying LICENSE file.
+ *
+ * This software is provided "AS IS" with no warranty of any kind,
+ * express or implied, and with no claim as to its suitability for any
+ * purpose.
+ *
+ */
+
+namespace lemon {
+/*!
+
+
+
+\page lgf-format Lemon Graph Format (LGF)
+
+The \e LGF is a <em>column oriented</em>
+file format for storing graphs and associated data like
+node and edge maps.
+
+Each line with \c '#' first non-whitespace
+character is considered as a comment line.
+
+Otherwise the file consists of sections starting with
+a header line. The header lines starts with an \c '@' character followed by the
+type of section. The standard section types are \c \@nodes, \c
+\@arcs and \c \@edges
+and \@attributes. Each header line may also have an optional
+\e name, which can be use to distinguish the sections of the same
+type.
+
+The standard sections are column oriented, each line consists of
+<em>token</em>s separated by whitespaces. A token can be \e plain or
+\e quoted. A plain token is just a sequence of non-whitespace characters,
+while a quoted token is a
+character sequence surrounded by double quotes, and it can also
+contain whitespaces and escape sequences. 
+
+The \c \@nodes section describes a set of nodes and associated
+maps. The first is a header line, it columns are the names of the
+maps appearing in the following lines.
+One of the maps must be called \c
+"label", which plays special role in the file.
+The following
+non-empty lines until the next section describes nodes of the
+graph. Each line contains the values of the node maps
+associated to the current node.
+
+\code
+ @nodes
+ label   coordinates size    title
+ 1       (10,20)     10      "First node"
+ 2       (80,80)     8       "Second node"
+ 3       (40,10)     10      "Third node"
+\endcode
+
+The \c \@arcs section is very similar to the \c \@nodes section,
+it again starts with a header line describing the names of the arc,
+but the \c "label" map is not obligatory here. The following lines
+describe the arcs. The first two tokens of each line are
+the source and the target node of the arc, respectively, then come the map
+values. The source and target tokens must be node labels.
+
+\code
+ @arcs
+ 	      capacity
+ 1   2   16
+ 1   3   12
+ 2   3   18
+\endcode
+
+The \c \@edges is just a synonym of \c \@arcs.
+
+The \c \@attributes section contains key-value pairs, each line
+consists of two tokens, an attribute name, and then an attribute value.
+
+\code
+ @attributes
+ source 1
+ target 3
+ caption "LEMON test digraph"
+\endcode
+
+*/
+}
+
+//  LocalWords:  whitespace whitespaces