diff --git a/errors.go b/errors.go
index 56db7c2..2affbab 100644
--- a/errors.go
+++ b/errors.go
@@ -7,4 +7,8 @@ var (
 	// closed. If encountered, the error should be considered terminal and
 	// retries will not be successful.
 	ErrSinkClosed = fmt.Errorf("events: sink closed")
+	// ErrNoSinks is returned if a write is issued to a round-robin sink without
+	// destiny. If encountered, destiny sinks should be added before calling
+	// Write again.
+	ErrNoSinks = fmt.Errorf("events: no destiny sink")
 )
diff --git a/roundrobin.go b/roundrobin.go
new file mode 100644
index 0000000..35e2caf
--- /dev/null
+++ b/roundrobin.go
@@ -0,0 +1,206 @@
+package events
+
+import (
+	"fmt"
+	"sync"
+
+	"github.com/sirupsen/logrus"
+)
+
+type eventRequest struct {
+	event Event
+	ch    chan error
+}
+
+// RoundRobin sends events to one of the Sinks in order. The goal of this
+// component is to dispatch events to configured endpoints. Reliability can be
+// provided by wrapping ingoing sinks.
+type RoundRobin struct {
+	sinks   []Sink
+	cursor  int
+	events  chan eventRequest
+	adds    chan configureRequest
+	removes chan configureRequest
+
+	shutdown chan struct{}
+	closed   chan struct{}
+	once     sync.Once
+}
+
+// NewRoundRobin appends one or more sinks to the list of sinks. The
+// round-robin behavior will be affected by the properties of the sink.
+// Generally, the sink should accept all messages and deal with reliability on
+// its own. Use of EventQueue and RetryingSink should be used here.
+func NewRoundRobin(sinks ...Sink) *RoundRobin {
+	b := RoundRobin{
+		sinks:    sinks,
+		events:   make(chan eventRequest),
+		adds:     make(chan configureRequest),
+		removes:  make(chan configureRequest),
+		shutdown: make(chan struct{}),
+		closed:   make(chan struct{}),
+	}
+
+	// Start the round-robin
+	go b.run()
+
+	return &b
+}
+
+// Write accepts an event to be dispatched to a sink.
+func (rr *RoundRobin) Write(event Event) error {
+	respChan := make(chan error, 1)
+	request := eventRequest{event: event, ch: respChan}
+	
+	select {
+	case rr.events <- request:
+	case <-rr.closed:
+		return ErrSinkClosed
+	}
+
+	select {
+	case err := <-respChan:
+		return err
+	case <-rr.closed:
+		return ErrSinkClosed
+	}
+}
+
+// Add the sink to the broadcaster.
+//
+// The provided sink must be comparable with equality. Typically, this just
+// works with a regular pointer type.
+func (rr *RoundRobin) Add(sink Sink) error {
+	return rr.configure(rr.adds, sink)
+}
+
+// Remove the provided sink.
+func (rr *RoundRobin) Remove(sink Sink) error {
+	return rr.configure(rr.removes, sink)
+}
+
+func (rr *RoundRobin) configure(ch chan configureRequest, sink Sink) error {
+	response := make(chan error, 1)
+
+	for {
+		select {
+		case ch <- configureRequest{
+			sink:     sink,
+			response: response}:
+			ch = nil
+		case err := <-response:
+			return err
+		case <-rr.closed:
+			return ErrSinkClosed
+		}
+	}
+}
+
+// Close the round-robin, ensuring that all messages are flushed to the
+// underlying sink before returning.
+func (rr *RoundRobin) Close() error {
+	rr.once.Do(func() {
+		close(rr.shutdown)
+	})
+
+	<-rr.closed
+	return nil
+}
+
+// run is the main round-robin loop, started when the round-robin is created.
+// Under normal conditions, it waits for events on the event channel. After
+// Close is called, this goroutine will exit.
+func (rr *RoundRobin) run() {
+	defer close(rr.closed)
+
+	for {
+		select {
+		case request := <-rr.events:
+			if len(rr.sinks) == 0 {
+				request.ch <- ErrNoSinks
+				break
+			}
+
+			rr.cursor++
+			rr.cursor %= len(rr.sinks)
+
+			for {
+				sink := rr.sinks[rr.cursor]
+
+				if err := sink.Write(request.event); err == ErrSinkClosed {
+					// remove closed sinks
+					rr.sinks = append(rr.sinks[:rr.cursor],
+						rr.sinks[rr.cursor+1:]...)
+					// check that it was not the only remaining sink
+					if len(rr.sinks) == 0 {
+						request.ch <- ErrNoSinks
+						break
+					}
+					// continue from the start if it was the last sink
+					rr.cursor %= len(rr.sinks)
+				} else {
+					request.ch <- err
+					break
+				}
+			}
+		case request := <-rr.adds:
+			// while we have to iterate for add/remove, common iteration for
+			// send is faster against slice.
+
+			var found bool
+			for _, sink := range rr.sinks {
+				if request.sink == sink {
+					found = true
+					break
+				}
+			}
+
+			if !found {
+				rr.sinks = append(rr.sinks, request.sink)
+			}
+			request.response <- nil
+		case request := <-rr.removes:
+			for i, sink := range rr.sinks {
+				if sink == request.sink {
+					rr.sinks = append(rr.sinks[:i], rr.sinks[i+1:]...)
+					if len(rr.sinks) == 0 {
+						rr.cursor = 0
+					} else if rr.cursor >= i {
+						// decrease the cursor if the remove sink was before
+						rr.cursor--
+						rr.cursor %= len(rr.sinks)
+					}
+					break
+				}
+			}
+			request.response <- nil
+		case <-rr.shutdown:
+			// close all the underlying sinks
+			for _, sink := range rr.sinks {
+				if err := sink.Close(); err != nil && err != ErrSinkClosed {
+					logrus.WithField("events.sink", sink).WithError(err).
+						Errorf("round-robin: closing sink failed")
+				}
+			}
+			return
+		}
+	}
+}
+
+func (rr *RoundRobin) String() string {
+	// Serialize copy of this round-robin without the sync.Once, to avoid
+	// a data race.
+
+	rr2 := map[string]interface{}{
+		"sinks":   rr.sinks,
+		"cursor":  rr.cursor,
+		"events":  rr.events,
+		"adds":    rr.adds,
+		"removes": rr.removes,
+
+		"shutdown": rr.shutdown,
+		"closed":   rr.closed,
+	}
+
+	return fmt.Sprint(rr2)
+}
diff --git a/roundrobin_test.go b/roundrobin_test.go
new file mode 100644
index 0000000..752262e
--- /dev/null
+++ b/roundrobin_test.go
@@ -0,0 +1,89 @@
+package events
+
+import (
+	"sync"
+	"testing"
+)
+
+func TestRoundRobin(t *testing.T) {
+	const nEvents = 1000
+	const nSinks  = 10
+	var sinks []Sink
+	rr := NewRoundRobin()
+	for i := 0; i < nSinks; i++ {
+		sinks = append(sinks, newTestSink(t, nEvents/nSinks))
+		rr.Add(sinks[i])
+		rr.Add(sinks[i]) // noop
+	}
+
+	var wg sync.WaitGroup
+	for i := 0; i < nEvents; i++ {
+		wg.Add(1)
+		go func(event Event) {
+			if err := rr.Write(event); err != nil {
+				t.Fatalf("error writing event %v: %v", event, err)
+			}
+			wg.Done()
+		}("event")
+	}
+
+	wg.Wait() // Wait until writes complete
+
+	for _, sink := range sinks {
+		rr.Remove(sink)
+	}
+
+	// sending one more should trigger no sink failure.
+	if err := rr.Write("one-more"); err != ErrNoSinks {
+		t.Fatalf("unexpected error: %v != %v", err, ErrNoSinks)
+	}
+
+	// add them back to test closing.
+	for i := range sinks {
+		rr.Add(sinks[i])
+	}
+
+	checkClose(t, rr)
+
+	// Iterate through the sinks and check that they all have the expected length.
+	for _, sink := range sinks {
+		ts := sink.(*testSink)
+		ts.mu.Lock()
+		defer ts.mu.Unlock()
+
+		if len(ts.events) != nEvents/nSinks {
+			t.Fatalf("not all events ended up in testsink: len(testSink) == %d, not %d", len(ts.events), nEvents/nSinks)
+		}
+
+		if !ts.closed {
+			t.Fatalf("sink should have been closed")
+		}
+	}
+}
+
+func BenchmarkRoundRobin10(b *testing.B) {
+	benchmarkRoundRobin(b, 10)
+}
+
+func BenchmarkRoundRobin100(b *testing.B) {
+	benchmarkRoundRobin(b, 100)
+}
+
+func BenchmarkRoundRobin1000(b *testing.B) {
+	benchmarkRoundRobin(b, 1000)
+}
+
+func BenchmarkRoundRobin10000(b *testing.B) {
+	benchmarkRoundRobin(b, 10000)
+}
+
+func benchmarkRoundRobin(b *testing.B, nsinks int) {
+	b.StopTimer()
+	var sinks []Sink
+	for i := 0; i < nsinks; i++ {
+		sinks = append(sinks, newTestSink(b, b.N/nsinks + 1))
+	}
+	b.StartTimer()
+
+	benchmarkSink(b, NewRoundRobin(sinks...))
+}