diff --git a/src/testing/benchmark.go b/src/testing/benchmark.go
index a5b163201e..b092a9d9e2 100644
--- a/src/testing/benchmark.go
+++ b/src/testing/benchmark.go
@@ -49,7 +49,7 @@ type B struct {
 	N                int
 	previousN        int           // number of iterations in the previous run
 	previousDuration time.Duration // total duration of the previous run
-	benchmark        InternalBenchmark
+	benchFunc        func(b *B)
 	bytes            int64
 	timerOn          bool
 	showAllocResult  bool
@@ -132,7 +132,7 @@ func (b *B) runN(n int) {
 	b.parallelism = 1
 	b.ResetTimer()
 	b.StartTimer()
-	b.benchmark.F(b)
+	b.benchFunc(b)
 	b.StopTimer()
 	b.previousN = n
 	b.previousDuration = b.duration
@@ -204,7 +204,7 @@ func (b *B) launch() {
 	// Signal that we're done whether we return normally
 	// or by FailNow's runtime.Goexit.
 	defer func() {
-		b.signal <- b
+		b.signal <- true
 	}()
 
 	b.runN(n)
@@ -339,9 +339,10 @@ func runBenchmarksInternal(matchString func(pat, str string) (bool, error), benc
 			runtime.GOMAXPROCS(procs)
 			b := &B{
 				common: common{
-					signal: make(chan interface{}),
+					signal: make(chan bool),
+					name:   Benchmark.Name,
 				},
-				benchmark: Benchmark,
+				benchFunc: Benchmark.F,
 			}
 			benchName := benchmarkName(Benchmark.Name, procs)
 			fmt.Printf("%-*s\t", maxlen, benchName)
@@ -476,9 +477,9 @@ func (b *B) SetParallelism(p int) {
 func Benchmark(f func(b *B)) BenchmarkResult {
 	b := &B{
 		common: common{
-			signal: make(chan interface{}),
+			signal: make(chan bool),
 		},
-		benchmark: InternalBenchmark{"", f},
+		benchFunc: f,
 	}
 	return b.run()
 }
diff --git a/src/testing/sub_test.go b/src/testing/sub_test.go
new file mode 100644
index 0000000000..8cb13ee101
--- /dev/null
+++ b/src/testing/sub_test.go
@@ -0,0 +1,101 @@
+// Copyright 2016 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.
+
+package testing
+
+func TestTestContext(t *T) {
+	const (
+		add1 = 0
+		done = 1
+	)
+	// After each of the calls are applied to the context, the
+	type call struct {
+		typ int // run or done
+		// result from applying the call
+		running int
+		waiting int
+		started bool
+	}
+	testCases := []struct {
+		max int
+		run []call
+	}{{
+		max: 1,
+		run: []call{
+			{typ: add1, running: 1, waiting: 0, started: true},
+			{typ: done, running: 0, waiting: 0, started: false},
+		},
+	}, {
+		max: 1,
+		run: []call{
+			{typ: add1, running: 1, waiting: 0, started: true},
+			{typ: add1, running: 1, waiting: 1, started: false},
+			{typ: done, running: 1, waiting: 0, started: true},
+			{typ: done, running: 0, waiting: 0, started: false},
+			{typ: add1, running: 1, waiting: 0, started: true},
+		},
+	}, {
+		max: 3,
+		run: []call{
+			{typ: add1, running: 1, waiting: 0, started: true},
+			{typ: add1, running: 2, waiting: 0, started: true},
+			{typ: add1, running: 3, waiting: 0, started: true},
+			{typ: add1, running: 3, waiting: 1, started: false},
+			{typ: add1, running: 3, waiting: 2, started: false},
+			{typ: add1, running: 3, waiting: 3, started: false},
+			{typ: done, running: 3, waiting: 2, started: true},
+			{typ: add1, running: 3, waiting: 3, started: false},
+			{typ: done, running: 3, waiting: 2, started: true},
+			{typ: done, running: 3, waiting: 1, started: true},
+			{typ: done, running: 3, waiting: 0, started: true},
+			{typ: done, running: 2, waiting: 0, started: false},
+			{typ: done, running: 1, waiting: 0, started: false},
+			{typ: done, running: 0, waiting: 0, started: false},
+		},
+	}}
+	for i, tc := range testCases {
+		ctx := &testContext{
+			startParallel: make(chan bool),
+			maxParallel:   tc.max,
+		}
+		for j, call := range tc.run {
+			doCall := func(f func()) chan bool {
+				done := make(chan bool)
+				go func() {
+					f()
+					done <- true
+				}()
+				return done
+			}
+			started := false
+			switch call.typ {
+			case add1:
+				signal := doCall(ctx.waitParallel)
+				select {
+				case <-signal:
+					started = true
+				case ctx.startParallel <- true:
+					<-signal
+				}
+			case done:
+				signal := doCall(ctx.release)
+				select {
+				case <-signal:
+				case <-ctx.startParallel:
+					started = true
+					<-signal
+				}
+			}
+			if started != call.started {
+				t.Errorf("%d:%d:started: got %v; want %v", i, j, started, call.started)
+			}
+			if ctx.running != call.running {
+				t.Errorf("%d:%d:running: got %v; want %v", i, j, ctx.running, call.running)
+			}
+			if ctx.numWaiting != call.waiting {
+				t.Errorf("%d:%d:waiting: got %v; want %v", i, j, ctx.numWaiting, call.waiting)
+			}
+		}
+	}
+}
diff --git a/src/testing/testing.go b/src/testing/testing.go
index 981883e07a..edd6c4fe74 100644
--- a/src/testing/testing.go
+++ b/src/testing/testing.go
@@ -147,6 +147,7 @@ import (
 	"bytes"
 	"flag"
 	"fmt"
+	"io"
 	"os"
 	"runtime"
 	"runtime/debug"
@@ -197,14 +198,18 @@ var (
 type common struct {
 	mu       sync.RWMutex // guards output and failed
 	output   []byte       // Output generated by test or benchmark.
+	w        io.Writer    // For flushToParent.
 	failed   bool         // Test or benchmark has failed.
 	skipped  bool         // Test of benchmark has been skipped.
 	finished bool
 
+	parent   *common
+	name     string    // Name of test or benchmark.
 	start    time.Time // Time test or benchmark started
 	duration time.Duration
-	self     interface{}      // To be sent on signal channel when done.
-	signal   chan interface{} // Output for serial tests.
+	barrier  chan bool // To signal parallel subtests they may start.
+	signal   chan bool // To signal a test is done.
+	sub      []*T      // Queue of subtests to be run in parallel.
 }
 
 // Short reports whether the -test.short flag is set.
@@ -251,6 +256,22 @@ func decorate(s string) string {
 	return buf.String()
 }
 
+// flushToParent writes c.output to the parent after first writing the header
+// with the given format and arguments.
+func (c *common) flushToParent(format string, args ...interface{}) {
+	p := c.parent
+	p.mu.Lock()
+	defer p.mu.Unlock()
+
+	fmt.Fprintf(p.w, format, args...)
+	fmt.Fprintln(p.w)
+
+	c.mu.Lock()
+	defer c.mu.Unlock()
+	io.Copy(p.w, bytes.NewReader(c.output))
+	c.output = c.output[:0]
+}
+
 // fmtDuration returns a string representing d in the form "87.00s".
 func fmtDuration(d time.Duration) string {
 	return fmt.Sprintf("%.2fs", d.Seconds())
@@ -293,15 +314,17 @@ var _ TB = (*B)(nil)
 // may be called simultaneously from multiple goroutines.
 type T struct {
 	common
-	name          string // Name of test.
-	isParallel    bool
-	startParallel chan bool // Parallel tests will wait on this.
+	isParallel bool
+	context    *testContext // For running tests and subtests.
 }
 
 func (c *common) private() {}
 
 // Fail marks the function as having failed but continues execution.
 func (c *common) Fail() {
+	if c.parent != nil {
+		c.parent.Fail()
+	}
 	c.mu.Lock()
 	defer c.mu.Unlock()
 	c.failed = true
@@ -437,8 +460,13 @@ func (t *T) Parallel() {
 	// in the test duration. Record the elapsed time thus far and reset the
 	// timer afterwards.
 	t.duration += time.Since(t.start)
-	t.signal <- (*T)(nil) // Release main testing loop
-	<-t.startParallel     // Wait for serial tests to finish
+
+	// Add to the list of tests to be released by the parent.
+	t.parent.sub = append(t.parent.sub, t)
+
+	t.signal <- true   // Release calling test.
+	<-t.parent.barrier // Wait for the parent test to complete.
+	t.context.waitParallel()
 	t.start = time.Now()
 }
 
@@ -449,8 +477,8 @@ type InternalTest struct {
 	F    func(*T)
 }
 
-func tRunner(t *T, test *InternalTest) {
-	// When this goroutine is done, either because test.F(t)
+func tRunner(t *T, fn func(t *T)) {
+	// When this goroutine is done, either because fn(t)
 	// returned normally or because a test failure triggered
 	// a call to runtime.Goexit, record the duration and send
 	// a signal saying that the test is done.
@@ -466,14 +494,87 @@ func tRunner(t *T, test *InternalTest) {
 			t.report()
 			panic(err)
 		}
-		t.signal <- t
+
+		if len(t.sub) > 0 {
+			// Run parallel subtests.
+			// Decrease the running count for this test.
+			t.context.release()
+			// Release the parallel subtests.
+			close(t.barrier)
+			// Wait for subtests to complete.
+			for _, sub := range t.sub {
+				<-sub.signal
+			}
+			if !t.isParallel {
+				// Reacquire the count for sequential tests. See comment in Run.
+				t.context.waitParallel()
+			}
+		} else if t.isParallel {
+			// Only release the count for this test if it was run as a parallel
+			// test. See comment in Run method.
+			t.context.release()
+		}
+		t.report() // Report after all subtests have finished.
+
+		t.signal <- true
 	}()
 
 	t.start = time.Now()
-	test.F(t)
+	fn(t)
 	t.finished = true
 }
 
+// testContext holds all fields that are common to all tests. This includes
+// synchronization primitives to run at most *parallel tests.
+type testContext struct {
+	mu sync.Mutex
+
+	// Channel used to signal tests that are ready to be run in parallel.
+	startParallel chan bool
+
+	// running is the number of tests currently running in parallel.
+	// This does not include tests that are waiting for subtests to complete.
+	running int
+
+	// numWaiting is the number tests waiting to be run in parallel.
+	numWaiting int
+
+	// maxParallel is a copy of the parallel flag.
+	maxParallel int
+}
+
+func newTestContext(maxParallel int) *testContext {
+	return &testContext{
+		startParallel: make(chan bool),
+		maxParallel:   *parallel,
+		running:       1, // Set the count to 1 for the main (sequential) test.
+	}
+}
+
+func (c *testContext) waitParallel() {
+	c.mu.Lock()
+	if c.running < c.maxParallel {
+		c.running++
+		c.mu.Unlock()
+		return
+	}
+	c.numWaiting++
+	c.mu.Unlock()
+	<-c.startParallel
+}
+
+func (c *testContext) release() {
+	c.mu.Lock()
+	if c.numWaiting == 0 {
+		c.running--
+		c.mu.Unlock()
+		return
+	}
+	c.numWaiting--
+	c.mu.Unlock()
+	c.startParallel <- true // Pick a waiting test to be run.
+}
+
 // An internal function but exported because it is cross-package; part of the implementation
 // of the "go test" command.
 func Main(matchString func(pat, str string) (bool, error), tests []InternalTest, benchmarks []InternalBenchmark, examples []InternalExample) {
@@ -526,15 +627,18 @@ func (m *M) Run() int {
 }
 
 func (t *T) report() {
+	if t.parent == nil {
+		return
+	}
 	dstr := fmtDuration(t.duration)
-	format := "--- %s: %s (%s)\n%s"
+	format := "--- %s: %s (%s)"
 	if t.Failed() {
-		fmt.Printf(format, "FAIL", t.name, dstr, t.output)
+		t.flushToParent(format, "FAIL", t.name, dstr)
 	} else if *chatty {
 		if t.Skipped() {
-			fmt.Printf(format, "SKIP", t.name, dstr, t.output)
+			t.flushToParent(format, "SKIP", t.name, dstr)
 		} else {
-			fmt.Printf(format, "PASS", t.name, dstr, t.output)
+			t.flushToParent(format, "PASS", t.name, dstr)
 		}
 	}
 }
@@ -547,63 +651,55 @@ func RunTests(matchString func(pat, str string) (bool, error), tests []InternalT
 	}
 	for _, procs := range cpuList {
 		runtime.GOMAXPROCS(procs)
-		// We build a new channel tree for each run of the loop.
-		// collector merges in one channel all the upstream signals from parallel tests.
-		// If all tests pump to the same channel, a bug can occur where a test
-		// kicks off a goroutine that Fails, yet the test still delivers a completion signal,
-		// which skews the counting.
-		var collector = make(chan interface{})
-
-		numParallel := 0
-		startParallel := make(chan bool)
-
-		for i := 0; i < len(tests); i++ {
-			matched, err := matchString(*match, tests[i].Name)
-			if err != nil {
-				fmt.Fprintf(os.Stderr, "testing: invalid regexp for -test.run: %s\n", err)
-				os.Exit(1)
-			}
-			if !matched {
-				continue
-			}
-			testName := tests[i].Name
-			t := &T{
-				common: common{
-					signal: make(chan interface{}),
-				},
-				name:          testName,
-				startParallel: startParallel,
-			}
-			t.self = t
-			if *chatty {
-				fmt.Printf("=== RUN   %s\n", t.name)
-			}
-			go tRunner(t, &tests[i])
-			out := (<-t.signal).(*T)
-			if out == nil { // Parallel run.
-				go func() {
-					collector <- <-t.signal
-				}()
-				numParallel++
-				continue
-			}
-			t.report()
-			ok = ok && !out.Failed()
+		ctx := newTestContext(*parallel)
+		t := &T{
+			common: common{
+				signal:  make(chan bool),
+				barrier: make(chan bool),
+				w:       os.Stdout,
+			},
+			context: ctx,
 		}
 
-		running := 0
-		for numParallel+running > 0 {
-			if running < *parallel && numParallel > 0 {
-				startParallel <- true
-				running++
-				numParallel--
-				continue
+		tRunner(t, func(t *T) {
+			for i := 0; i < len(tests); i++ {
+				// TODO: a version of this will be the Run method.
+				matched, err := matchString(*match, tests[i].Name)
+				if err != nil {
+					fmt.Fprintf(os.Stderr, "testing: invalid regexp for -test.run: %s\n", err)
+					os.Exit(1)
+				}
+				if !matched {
+					continue
+				}
+				testName := tests[i].Name
+				t := &T{
+					common: common{
+						barrier: make(chan bool),
+						signal:  make(chan bool),
+						name:    testName,
+						parent:  &t.common,
+					},
+					context: t.context,
+				}
+
+				if *chatty {
+					fmt.Printf("=== RUN   %s\n", t.name)
+				}
+				// Instead of reducing the running count of this test before calling the
+				// tRunner and increasing it afterwards, we rely on tRunner keeping the
+				// count correct. This ensures that a sequence of sequential tests runs
+				// without being preempted, even when their parent is a parallel test. This
+				// may especially reduce surprises if *parallel == 1.
+				go tRunner(t, tests[i].F)
+				<-t.signal
 			}
-			t := (<-collector).(*T)
-			t.report()
-			ok = ok && !t.Failed()
-			running--
-		}
+			// Run catching the signal rather than the tRunner as a separate
+			// goroutine to avoid adding a goroutine during the sequential
+			// phase as this pollutes the stacktrace output when aborting.
+			go func() { <-t.signal }()
+		})
+		ok = ok && !t.Failed()
 	}
 	return
 }
diff --git a/test/fixedbugs/bug229.go b/test/fixedbugs/bug229.go
index 19776881d1..5cc3988555 100644
--- a/test/fixedbugs/bug229.go
+++ b/test/fixedbugs/bug229.go
@@ -14,7 +14,7 @@ func main() {
 	// make sure error mentions that
 	// name is unexported, not just "name not found".
 
-	t.name = nil	// ERROR "unexported"
+	t.common.name = nil	// ERROR "unexported"
 	
 	println(testing.anyLowercaseName("asdf"))	// ERROR "unexported" "undefined: testing.anyLowercaseName"
 }