diff --git a/classify.go b/classify.go
index 9ce6e04..ffded99 100644
--- a/classify.go
+++ b/classify.go
@@ -1,8 +1,10 @@
 package sisyphus
 
 import (
+	"math/rand"
 	"os"
 	"path/filepath"
+	"time"
 
 	log "github.com/sirupsen/logrus"
 
@@ -202,6 +204,26 @@ func (m *Mail) Classify(db *bolt.DB, dir Maildir) (err error) {
 func Junk(db *bolt.DB, wordlist []string) (junk bool, prob float64, err error) {
 	var probabilities []float64
 
+	// If the wordlist is too long, let us only select a random sample
+	// for analysis. This prevents cheating by adding lots of good text
+	// to a Junk mail
+	if len(wordlist) > 50 {
+		wordlistTemp := make(map[string]interface{})
+
+		rand.Seed(time.Now().UnixNano())
+
+		for len(wordlistTemp) < 50 {
+			wordlistTemp[wordlist[rand.Intn(len(wordlist)-1)]] = nil
+
+		}
+
+		var wordlistTempSlice []string
+		for key, _ := range wordlistTemp {
+			wordlistTempSlice = append(wordlistTempSlice, key)
+		}
+		wordlist = wordlistTempSlice
+	}
+
 	// initial value should be no junk
 	prob = 1.0
 
diff --git a/classify_test.go b/classify_test.go
index 3002310..90d4116 100644
--- a/classify_test.go
+++ b/classify_test.go
@@ -124,4 +124,97 @@ var _ = Describe("Classify Mails", func() {
 
 		})
 	})
+
+	Context("Only classify a random subset of the words in overly long mails", func() {
+		BeforeEach(func() {
+			// Load empty Maildir2
+			err = LoadMaildirs([]Maildir{
+				"test/Maildir2",
+			})
+			Ω(err).ShouldNot(HaveOccurred())
+
+			// Load db
+			dbs, err = LoadDatabases([]Maildir{
+				"test/Maildir2",
+			})
+			Ω(err).ShouldNot(HaveOccurred())
+
+		})
+		AfterEach(func() {
+			// Cleanup
+			CloseDatabases(dbs)
+
+			err = os.RemoveAll("test/Maildir2")
+			Ω(err).ShouldNot(HaveOccurred())
+		})
+
+		It("learned nothing and thus return always good", func() {
+
+			_, _, err := Junk(dbs["test/Maildir2"], []string{
+				"Carlo",
+				"0",
+				"1",
+				"2",
+				"3",
+				"4",
+				"5",
+				"6",
+				"7",
+				"8",
+				"9",
+				"10",
+				"11",
+				"12",
+				"13",
+				"14",
+				"15",
+				"16",
+				"17",
+				"18",
+				"19",
+				"20",
+				"21",
+				"22",
+				"23",
+				"24",
+				"25",
+				"26",
+				"27",
+				"28",
+				"29",
+				"30",
+				"31",
+				"32",
+				"33",
+				"34",
+				"35",
+				"36",
+				"37",
+				"38",
+				"39",
+				"40",
+				"41",
+				"42",
+				"43",
+				"44",
+				"45",
+				"46",
+				"47",
+				"48",
+				"49",
+				"50",
+				"51",
+				"52",
+				"53",
+				"54",
+				"55",
+				"56",
+				"57",
+				"58",
+				"59",
+			})
+
+			Ω(err).ShouldNot(HaveOccurred())
+		})
+	})
 })
diff --git a/mail.go b/mail.go
index 3489d4d..ec88157 100644
--- a/mail.go
+++ b/mail.go
@@ -225,18 +225,14 @@ func wordlist(s string) (l []string, err error) {
 		}
 	}
 
-	// only the first 200 words count
-	maxWords := int(math.Min(200, float64(len(clean))))
+	// only the first 1000 words count
+	maxWords := int(math.Min(1000, float64(len(clean))))
 	for i := 0; i < maxWords; i++ {
 		w := clean[i]
 		list[w]++
 	}
 
-	for word, count := range list {
-		if count > 10 {
-			continue
-		}
-
+	for word := range list {
 		l = append(l, word)
 	}