diff --git a/src/pkg/exp/locale/collate/regtest.go b/src/pkg/exp/locale/collate/regtest.go
new file mode 100644
index 00000000000..6a2891c7504
--- /dev/null
+++ b/src/pkg/exp/locale/collate/regtest.go
@@ -0,0 +1,180 @@
+// Copyright 2012 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// +build ignore
+
+package main
+
+import (
+	"archive/zip"
+	"bufio"
+	"bytes"
+	"exp/locale/collate"
+	"flag"
+	"fmt"
+	"io"
+	"io/ioutil"
+	"log"
+	"net/http"
+	"os"
+	"path"
+	"regexp"
+	"strconv"
+	"strings"
+	"unicode"
+)
+
+// This regression test runs tests for the test files in CollationTest.zip
+// (taken from http://www.unicode.org/Public/UCA/<unicode.Version>/).
+//
+// The test files have the following form:
+// # header
+// 0009 0021;	# ('\u0009') <CHARACTER TABULATION>	[| | | 0201 025E]
+// 0009 003F;	# ('\u0009') <CHARACTER TABULATION>	[| | | 0201 0263]
+// 000A 0021;	# ('\u000A') <LINE FEED (LF)>	[| | | 0202 025E]
+// 000A 003F;	# ('\u000A') <LINE FEED (LF)>	[| | | 0202 0263]
+//
+// The part before the semicolon is the hex representation of a sequence
+// of runes. After the hash mark is a comment. The strings
+// represented by rune sequence are in the file in sorted order, as
+// defined by the DUCET.
+
+var url = flag.String("url",
+	"http://www.unicode.org/Public/UCA/"+unicode.Version+"/CollationTest.zip",
+	"URL of Unicode collation tests zip file")
+var localFiles = flag.Bool("local",
+	false,
+	"data files have been copied to the current directory; for debugging only")
+
+type Test struct {
+	name    string
+	str     []string
+	comment []string
+}
+
+var versionRe = regexp.MustCompile(`# UCA Version: (.*)\n?$`)
+var testRe = regexp.MustCompile(`^([\dA-F ]+);.*# (.*)\n?$`)
+
+func Error(e error) {
+	if e != nil {
+		log.Fatal(e)
+	}
+}
+
+func loadTestData() []Test {
+	if *localFiles {
+		pwd, _ := os.Getwd()
+		*url = "file://" + path.Join(pwd, path.Base(*url))
+	}
+	t := &http.Transport{}
+	t.RegisterProtocol("file", http.NewFileTransport(http.Dir("/")))
+	c := &http.Client{Transport: t}
+	resp, err := c.Get(*url)
+	Error(err)
+	if resp.StatusCode != 200 {
+		log.Fatalf(`bad GET status for "%s": %s`, *url, resp.Status)
+	}
+	f := resp.Body
+	buffer, err := ioutil.ReadAll(f)
+	f.Close()
+	Error(err)
+	archive, err := zip.NewReader(bytes.NewReader(buffer), int64(len(buffer)))
+	Error(err)
+	tests := []Test{}
+	for _, f := range archive.File {
+		// Skip the short versions, which are simply duplicates of the long versions.
+		if strings.Contains(f.Name, "SHORT") || f.FileInfo().IsDir() {
+			continue
+		}
+		ff, err := f.Open()
+		Error(err)
+		defer ff.Close()
+		input := bufio.NewReader(ff)
+		test := Test{name: path.Base(f.Name)}
+		for {
+			line, err := input.ReadString('\n')
+			if err != nil {
+				if err == io.EOF {
+					break
+				}
+				log.Fatal(err)
+			}
+			if len(line) <= 1 || line[0] == '#' {
+				if m := versionRe.FindStringSubmatch(line); m != nil {
+					if m[1] != unicode.Version {
+						log.Printf("warning:%s: version is %s; want %s", f.Name, m[1], unicode.Version)
+					}
+				}
+				continue
+			}
+			m := testRe.FindStringSubmatch(line)
+			if m == nil || len(m) < 3 {
+				log.Fatalf(`Failed to parse: "%s" result: %#v`, line, m)
+			}
+			str := ""
+			for _, split := range strings.Split(m[1], " ") {
+				r, err := strconv.ParseUint(split, 16, 64)
+				Error(err)
+				str += string(rune(r))
+			}
+			test.str = append(test.str, str)
+			test.comment = append(test.comment, m[2])
+		}
+		tests = append(tests, test)
+	}
+	return tests
+}
+
+var errorCount int
+
+func fail(t Test, pattern string, args ...interface{}) {
+	format := fmt.Sprintf("error:%s:%s", t.name, pattern)
+	log.Printf(format, args...)
+	errorCount++
+	if errorCount > 30 {
+		log.Fatal("too many errors")
+	}
+}
+
+func runes(b []byte) []rune {
+	return []rune(string(b))
+}
+
+func doTest(t Test) {
+	c := collate.Root
+	c.Strength = collate.Tertiary
+	b := &collate.Buffer{}
+	if strings.Contains(t.name, "NON_IGNOR") {
+		c.Alternate = collate.AltNonIgnorable
+	}
+
+	prev := []byte(t.str[0])
+	for i := 1; i < len(t.str); i++ {
+		s := []byte(t.str[i])
+		ka := c.Key(b, prev)
+		kb := c.Key(b, s)
+		if r := bytes.Compare(ka, kb); r == 1 {
+			fail(t, "%d: Key(%.4X) < Key(%.4X) (%X < %X) == %d; want -1 or 0", i, runes(prev), runes(s), ka, kb, r)
+			prev = s
+			continue
+		}
+		if r := c.Compare(b, prev, s); r == 1 {
+			fail(t, "%d: Compare(%.4X, %.4X) == %d; want -1 or 0", i, runes(prev), runes(s), r)
+		}
+		if r := c.Compare(b, s, prev); r == -1 {
+			fail(t, "%d: Compare(%.4X, %.4X) == %d; want 1 or 0", i, runes(s), runes(prev), r)
+		}
+		prev = s
+	}
+}
+
+func main() {
+	flag.Parse()
+	for _, test := range loadTestData() {
+		doTest(test)
+	}
+	if errorCount == 0 {
+		fmt.Println("PASS")
+	}
+}