diff --git a/README.md b/README.md
index 32ee0b7d76..e16e165208 100644
--- a/README.md
+++ b/README.md
@@ -155,6 +155,10 @@ simply visit all statements and expressions and not produce a
 meaningful state. (An example of a use of state it to track scope at
 each point in the tree.)
 
+**ancestor**`(node, visitors, base, state)` does a 'simple' walk over
+a tree, building up an array of ancestor nodes (including the current node)
+and passing the array to callbacks in the `state` parameter.
+
 **recursive**`(node, state, functions, base)` does a 'recursive'
 walk, where the walker functions are responsible for continuing the
 walk on the child nodes of their target node. `state` is the start
diff --git a/util/walk.js b/util/walk.js
index 4c0d9a7f88..f8c697c8c4 100644
--- a/util/walk.js
+++ b/util/walk.js
@@ -32,6 +32,23 @@
     c(node, state);
   };
 
+  // An ancestor walk builds up an array of ancestor nodes (including
+  // the current node) and passes them to the callback as the state parameter.
+  exports.ancestor = function(node, visitors, base, state) {
+    if (!base) base = exports.base;
+    if (!state) state = [];
+    function c(node, st, override) {
+      var type = override || node.type, found = visitors[type];
+      if (node != st[st.length - 1]) {
+        st = st.slice();
+        st.push(node);
+      }
+      base[type](node, st, c);
+      if (found) found(node, st);
+    }
+    c(node, state);
+  };
+
   // A recursive walk is one where your functions override the default
   // walkers. They can modify and replace the state parameter that's
   // threaded through the walk, and can opt how and whether to walk