diff --git a/include/oxli/hashgraph.hh b/include/oxli/hashgraph.hh
index f450a42e6b..f0d0d4e2db 100644
--- a/include/oxli/hashgraph.hh
+++ b/include/oxli/hashgraph.hh
@@ -293,6 +293,7 @@ public:
         : Hashgraph(ksize, new BitStorage(sizes)) { } ;
 
     void update_from(const Nodegraph &other);
+    double similarity(const Nodegraph &other);
 };
 
 }
diff --git a/include/oxli/storage.hh b/include/oxli/storage.hh
index 33fb6f7f73..37a69d9df8 100644
--- a/include/oxli/storage.hh
+++ b/include/oxli/storage.hh
@@ -226,6 +226,7 @@ public:
     }
 
     void update_from(const BitStorage&);
+    double similarity(const BitStorage&);
 };
 
 
diff --git a/khmer/_oxli/graphs.pxd b/khmer/_oxli/graphs.pxd
index ce4d290e86..16e8627591 100644
--- a/khmer/_oxli/graphs.pxd
+++ b/khmer/_oxli/graphs.pxd
@@ -204,6 +204,7 @@ cdef extern from "oxli/hashgraph.hh" namespace "oxli" nogil:
         CpNodegraph(WordLength, vector[uint64_t])
 
         void update_from(const CpNodegraph &) except +oxli_raise_py_error
+        double similarity(const CpNodegraph &) except +oxli_raise_py_error
 
 
 cdef extern from "oxli/labelhash.hh" namespace "oxli":
diff --git a/khmer/_oxli/graphs.pyx b/khmer/_oxli/graphs.pyx
index 992ae526ae..bf9c9b4f5e 100644
--- a/khmer/_oxli/graphs.pyx
+++ b/khmer/_oxli/graphs.pyx
@@ -866,3 +866,6 @@ cdef class Nodegraph(Hashgraph):
 
     def update(self, Nodegraph other):
         deref(self._ng_this).update_from(deref(other._ng_this))
+
+    def similarity(self, Nodegraph other):
+        return deref(self._ng_this).similarity(deref(other._ng_this))
diff --git a/src/oxli/hashgraph.cc b/src/oxli/hashgraph.cc
index c9cd78d860..cc8aa0a3fe 100644
--- a/src/oxli/hashgraph.cc
+++ b/src/oxli/hashgraph.cc
@@ -906,6 +906,23 @@ void Nodegraph::update_from(const Nodegraph &otherBASE)
     }
 }
 
+double Nodegraph::similarity(const Nodegraph &otherBASE)
+{
+    if (_ksize != otherBASE._ksize) {
+        throw oxli_exception("both nodegraphs must have same k size");
+    }
+    BitStorage * myself = dynamic_cast<BitStorage *>(this->store);
+    const BitStorage * other;
+    other = dynamic_cast<const BitStorage*>(otherBASE.store);
+
+    // if dynamic_cast worked, then the pointers will be not null.
+    if (myself && other) {
+        return myself->similarity(*other);
+    } else {
+        throw oxli_exception("similarity failed with incompatible objects");
+    }
+}
+
 template void Hashgraph::consume_seqfile_and_tag<read_parsers::FastxReader>(
     std::string const &filename,
     unsigned int &total_reads,
diff --git a/src/oxli/storage.cc b/src/oxli/storage.cc
index 923843f451..8e9f559b20 100644
--- a/src/oxli/storage.cc
+++ b/src/oxli/storage.cc
@@ -95,6 +95,33 @@ void BitStorage::update_from(const BitStorage& other)
     }
 }
 
+double BitStorage::similarity(const BitStorage& other)
+{
+    if (_tablesizes != other._tablesizes) {
+        throw oxli_exception("both nodegraphs must have same table sizes");
+    }
+
+    uint64_t intersection = 0;
+    uint64_t union_size = 0;
+    for (unsigned int table_num = 0; table_num < _n_tables; table_num++) {
+        Byte * me = _counts[table_num];
+        Byte * ot = other._counts[table_num];
+        uint64_t tablesize = _tablesizes[table_num];
+        uint64_t tablebytes = tablesize / 8 + 1;
+
+        for (uint64_t index = 0; index < tablebytes; index++) {
+            // First, get how many values in common we have
+            intersection += __builtin_popcountll(me[index] & ot[index]);
+            union_size += __builtin_popcountll(me[index] | ot[index]);
+        }
+    }
+
+    if (union_size == 0) {
+        union_size = 1;
+    }
+
+    return double(intersection) / double(union_size);
+}
 
 void BitStorage::save(std::string outfilename, WordLength ksize)
 {
diff --git a/tests/test_nodegraph.py b/tests/test_nodegraph.py
index 607d521bfe..a5d1d42a7b 100755
--- a/tests/test_nodegraph.py
+++ b/tests/test_nodegraph.py
@@ -216,6 +216,26 @@ def test_update_from_diff_num_tables():
         print(str(err))
 
 
+def test_similarity_1():
+    nodegraph = khmer.Nodegraph(5, 1000, 4)
+    other_nodegraph = khmer.Nodegraph(5, 1000, 4)
+
+    assert nodegraph.similarity(other_nodegraph) == 0
+
+    other_nodegraph.count('AAAAA')
+
+    assert nodegraph.similarity(other_nodegraph) == 0
+
+    nodegraph.count('GCGCG')
+
+    assert nodegraph.similarity(other_nodegraph) == 0
+
+    nodegraph.count('AAAAA')
+    other_nodegraph.count('GCGCG')
+
+    assert nodegraph.similarity(other_nodegraph) == 1
+
+
 def test_n_occupied_1():
     filename = utils.get_test_data('random-20-a.fa')