diff --git a/src/main/scala/nak/cluster/GDBSCAN.scala b/src/main/scala/nak/cluster/GDBSCAN.scala
index be602ec..7332163 100644
--- a/src/main/scala/nak/cluster/GDBSCAN.scala
+++ b/src/main/scala/nak/cluster/GDBSCAN.scala
@@ -1,5 +1,7 @@
 package nak.cluster
 
+import scala.language.{ implicitConversions, postfixOps }
+
 import breeze.numerics._
 import breeze.linalg._
 import breeze.util._
@@ -26,13 +28,14 @@ class GDBSCAN[T](
    * @param data - each row is treated as a feature vector
    * @return clusters - a list of clusters with
    */
-  def cluster(data: DenseMatrix[T]): Seq[Cluster[T]] = {
+  def cluster(data: DenseMatrix[T]): Seq[AbstractCluster[T]] = {
     // Visited - using row indices
     val visited = MutableSet[Point[T]]()
     val clustered = MutableSet[Point[T]]()
 
     // Init points
     val points = for (row <- 0 until data.rows) yield Point(row)(data(row, ::).inner)
+    val noise = Noise[T]()
 
     // Start clustering
     points.collect {
@@ -44,12 +47,13 @@ class GDBSCAN[T](
           expand(point, neighbours, cluster)(points, visited, clustered)
           Some(cluster)
         } else {
+          noise add point
           None // noise
         }
-    }.flatten // remove noise
+    }.flatten :+ noise
   }
 
-  private def expand(point: Point[T], neighbours: Seq[Point[T]], cluster: Cluster[T])(implicit points: Seq[Point[T]], visited: MutableSet[Point[T]], clustered: MutableSet[Point[T]]) {
+  private def expand(point: Point[T], neighbours: Seq[Point[T]], cluster: AbstractCluster[T])(points: Seq[Point[T]], visited: MutableSet[Point[T]], clustered: MutableSet[Point[T]]) {
     cluster add point
     clustered add point
     neighbours.foldLeft(neighbours) {
@@ -80,19 +84,25 @@ object GDBSCAN {
     override def toString() = s"[$row]: $value"
   }
 
-  /** Cluster description */
-  case class Cluster[T](id: Long) {
-    private var _points = ListBuffer[Point[T]]()
-
+  abstract class AbstractCluster[T] {
+    val id: Long
+    
+    protected var _points = ListBuffer[Point[T]]()
+    
     def add(p: Point[T]) {
       _points += p
     }
-
+    
     def points: Seq[Point[T]] = Seq(_points: _*)
+  }
 
+  /** Cluster description */
+  case class Cluster[T](id: Long) extends AbstractCluster[T] {
     override def toString() = s"Cluster [$id]\t:\t${_points.size} points\t${_points mkString "|"}"
   }
 
+  case class Noise[T](id: Long = -1) extends AbstractCluster[T]
+
 }
 
 /**
@@ -122,4 +132,4 @@ object DBSCAN {
   def isCorePoint(minPoints: Double)(point: Point[Double], neighbours: Seq[Point[Double]]): Boolean = {
     neighbours.size >= minPoints
   }
-}
\ No newline at end of file
+}
diff --git a/src/test/scala/nak/cluster/GDBSCANTest.scala b/src/test/scala/nak/cluster/GDBSCANTest.scala
index fa200ef..c7715c5 100644
--- a/src/test/scala/nak/cluster/GDBSCANTest.scala
+++ b/src/test/scala/nak/cluster/GDBSCANTest.scala
@@ -28,9 +28,10 @@ class GDBSCANTest extends FlatSpec with Matchers {
     val cluster = gdbscan cluster input
     val clusterPoints = cluster.map(_.points.map(_.value.toArray))
 
-    cluster.size shouldBe 2
+    cluster.size shouldBe 3
     clusterPoints(0) should contain only (Array(0.9, 1.0), Array(1.0, 1.0), Array(1.0, 1.1))
     clusterPoints(1) should contain only (Array(15.0, 15.0), Array(15.0, 14.1), Array(15.3, 15.0))
+    clusterPoints(2) should contain only (Array(5.0,5.0))
   }
 
   it should "work with custom predicates" in {
@@ -55,8 +56,9 @@ class GDBSCANTest extends FlatSpec with Matchers {
     val cluster = gdbscan cluster input
     val clusterPoints = cluster.map(_.points.map(_.value.toArray))
 
-    cluster.size shouldBe 2
+    cluster.size shouldBe 3
     clusterPoints(0) should contain only (Array(1.0), Array(3.0))
     clusterPoints(1) should contain only (Array(2.0), Array(4.0))
+    clusterPoints(2) shouldBe empty
   }
-}
\ No newline at end of file
+}