diff --git a/events/loop/loop.go b/events/loop/loop.go
index bbc3b9c8..cdc2e07b 100644
--- a/events/loop/loop.go
+++ b/events/loop/loop.go
@@ -30,20 +30,24 @@ type Interface[T any] interface {
 }
 
 type loop[T any] struct {
-	queue   chan T
+	head *queueSegment[T]
+	tail *queueSegment[T]
+
+	segSize uint64
+
 	handler Handler[T]
 
 	closed  bool
 	closeCh chan struct{}
-	lock    sync.RWMutex
+
+	lock sync.Mutex
+
+	segPool sync.Pool
 }
 
 func New[T any](h Handler[T], size uint64) Interface[T] {
-	return &loop[T]{
-		queue:   make(chan T, size),
-		handler: h,
-		closeCh: make(chan struct{}),
-	}
+	l := new(loop[T])
+	return l.Reset(h, size)
 }
 
 func Empty[T any]() Interface[T] {
@@ -53,41 +57,88 @@ func Empty[T any]() Interface[T] {
 func (l *loop[T]) Run(ctx context.Context) error {
 	defer close(l.closeCh)
 
-	for {
-		req, ok := <-l.queue
-		if !ok {
-			return nil
+	current := l.head
+	for current != nil {
+		// Drain this segment in order. The channel will be closed either:
+		//   - when we "roll over" to a new segment, or
+		//   - when Close() is called for the final segment.
+		for req := range current.ch {
+			if err := l.handler.Handle(ctx, req); err != nil {
+				return err
+			}
 		}
 
-		if err := l.handler.Handle(ctx, req); err != nil {
-			return err
-		}
+		// Move to the next segment, and return this one to the pool.
+		next := current.next
+		l.putSegment(current)
+		current = next
 	}
+
+	return nil
 }
 
 func (l *loop[T]) Enqueue(req T) {
-	l.lock.RLock()
-	defer l.lock.RUnlock()
+	l.lock.Lock()
+	defer l.lock.Unlock()
 
 	if l.closed {
 		return
 	}
 
+	if l.tail == nil {
+		seg := l.getSegment()
+		l.head = seg
+		l.tail = seg
+	}
+
+	// First try to send to the current tail segment without blocking.
 	select {
-	case l.queue <- req:
-	case <-l.closeCh:
+	case l.tail.ch <- req:
+		return
+	default:
+		// Tail is full: create a new segment, link it, close the old tail, and
+		// send into the new tail.
+		newSeg := l.getSegment()
+		l.tail.next = newSeg
+		close(l.tail.ch)
+		l.tail = newSeg
+		l.tail.ch <- req
 	}
 }
 
 func (l *loop[T]) Close(req T) {
 	l.lock.Lock()
+	if l.closed {
+		// Already closed; just unlock and wait for Run to finish.
+		l.lock.Unlock()
+		<-l.closeCh
+		return
+	}
 	l.closed = true
+
+	// Ensure at least one segment exists.
+	if l.tail == nil {
+		seg := l.getSegment()
+		l.head = seg
+		l.tail = seg
+	}
+
+	// Enqueue the final request; if the tail is full, roll over as in Enqueue.
 	select {
-	case l.queue <- req:
-	case <-l.closeCh:
+	case l.tail.ch <- req:
+	default:
+		newSeg := l.getSegment()
+		l.tail.next = newSeg
+		close(l.tail.ch)
+		l.tail = newSeg
+		l.tail.ch <- req
 	}
-	close(l.queue)
+
+	// No more items will be enqueued; close the tail to signal completion.
+	close(l.tail.ch)
 	l.lock.Unlock()
+
+	// Wait for Run to finish draining everything.
 	<-l.closeCh
 }
 
@@ -102,10 +153,16 @@ func (l *loop[T]) Reset(h Handler[T], size uint64) Interface[T] {
 	l.closed = false
 	l.closeCh = make(chan struct{})
 	l.handler = h
+	l.segSize = size
+
+	// Initialize pool for this instantiation of T.
+	l.segPool.New = func() any {
+		return new(queueSegment[T])
+	}
 
-	// TODO: @joshvanl: use a ring buffer so that we don't need to reallocate and
-	// improve performance.
-	l.queue = make(chan T, size)
+	seg := l.getSegment()
+	l.head = seg
+	l.tail = seg
 
 	return l
 }
diff --git a/events/loop/loop_test.go b/events/loop/loop_test.go
new file mode 100644
index 00000000..ade3f336
--- /dev/null
+++ b/events/loop/loop_test.go
@@ -0,0 +1,221 @@
+/*
+Copyright 2025 The Dapr Authors
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+    http://www.apache.org/licenses/LICENSE-2.0
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+package loop
+
+import (
+	"context"
+	"sync"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+// testHandler is a simple handler that records all values it sees.
+type testHandler[T any] struct {
+	mu    sync.Mutex
+	seen  []T
+	err   error
+	delay time.Duration
+}
+
+func (h *testHandler[T]) Handle(ctx context.Context, v T) error {
+	if h.delay > 0 {
+		select {
+		case <-time.After(h.delay):
+		case <-ctx.Done():
+			return ctx.Err()
+		}
+	}
+
+	h.mu.Lock()
+	defer h.mu.Unlock()
+	h.seen = append(h.seen, v)
+	return h.err
+}
+
+func (h *testHandler[T]) Values() []T {
+	h.mu.Lock()
+	defer h.mu.Unlock()
+	out := make([]T, len(h.seen))
+	copy(out, h.seen)
+	return out
+}
+
+func TestLoop_EnqueueAndRunOrder_Unbounded(t *testing.T) {
+	ctx, cancel := context.WithCancel(t.Context())
+	defer cancel()
+
+	h := &testHandler[int]{}
+	const segmentSize = 4
+
+	l := New[int](h, segmentSize)
+
+	var wg sync.WaitGroup
+	wg.Add(1)
+	errCh := make(chan error, 1)
+	t.Cleanup(func() {
+		require.NoError(t, <-errCh)
+	})
+	go func() {
+		defer wg.Done()
+		errCh <- l.Run(ctx)
+	}()
+
+	// Enqueue more items than a single segment to force multiple channels.
+	const n = 25
+	for i := range n {
+		l.Enqueue(i)
+	}
+
+	// Close with a sentinel value so we can verify it is the last element.
+	const final = 999
+	l.Close(final)
+
+	wg.Wait()
+
+	got := h.Values()
+	require.Len(t, got, n+1, "handler should see all enqueued items plus final close item")
+
+	// First n values should be 0..n-1 in order.
+	for i := range n {
+		assert.Equal(t, i, got[i], "item at index %d out of order", i)
+	}
+
+	// Last one is the final close value.
+	assert.Equal(t, final, got[len(got)-1], "last item should be the Close() value")
+}
+
+func TestLoop_CloseTwiceIsSafe(t *testing.T) {
+	ctx, cancel := context.WithCancel(t.Context())
+	defer cancel()
+
+	h := &testHandler[int]{}
+	l := New[int](h, 2)
+
+	var wg sync.WaitGroup
+	wg.Add(1)
+	errCh := make(chan error, 1)
+	t.Cleanup(func() {
+		require.NoError(t, <-errCh)
+	})
+	go func() {
+		defer wg.Done()
+		errCh <- l.Run(ctx)
+	}()
+
+	l.Enqueue(1)
+	l.Close(2)
+
+	// Second close should not panic or deadlock.
+	done := make(chan struct{})
+	go func() {
+		l.Close(3)
+		close(done)
+	}()
+
+	select {
+	case <-done:
+	case <-time.After(time.Second):
+		require.Fail(t, "second Close should not block")
+	}
+
+	wg.Wait()
+
+	got := h.Values()
+	// First close enqueues 2, second close should be ignored.
+	assert.Contains(t, got, 1)
+	assert.Contains(t, got, 2)
+	assert.NotContains(t, got, 3)
+}
+
+func TestLoop_Reset(t *testing.T) {
+	ctx, cancel := context.WithCancel(t.Context())
+	defer cancel()
+
+	h1 := &testHandler[int]{}
+	l := New[int](h1, 2)
+
+	var wg1 sync.WaitGroup
+	wg1.Add(1)
+	errCh := make(chan error, 1)
+	t.Cleanup(func() {
+		require.NoError(t, <-errCh)
+	})
+	go func() {
+		defer wg1.Done()
+		errCh <- l.Run(ctx)
+	}()
+
+	l.Enqueue(1)
+	l.Close(2)
+	wg1.Wait()
+
+	assert.ElementsMatch(t, []int{1, 2}, h1.Values())
+
+	// Reset to a new handler and buffer size.
+	h2 := &testHandler[int]{}
+	l = l.Reset(h2, 8)
+
+	require.NotNil(t, l)
+
+	var wg2 sync.WaitGroup
+	wg2.Add(1)
+	errCh2 := make(chan error, 1)
+	t.Cleanup(func() {
+		require.NoError(t, <-errCh2)
+	})
+	go func() {
+		defer wg2.Done()
+		errCh2 <- l.Run(ctx)
+	}()
+
+	l.Enqueue(10)
+	l.Enqueue(11)
+	l.Close(12)
+	wg2.Wait()
+
+	assert.ElementsMatch(t, []int{10, 11, 12}, h2.Values())
+}
+
+func TestLoop_EnqueueAfterCloseIsDropped(t *testing.T) {
+	ctx, cancel := context.WithCancel(t.Context())
+	defer cancel()
+
+	h := &testHandler[int]{}
+	l := New[int](h, 2)
+
+	var wg sync.WaitGroup
+	wg.Add(1)
+	errCh := make(chan error, 1)
+	t.Cleanup(func() {
+		require.NoError(t, <-errCh)
+	})
+	go func() {
+		defer wg.Done()
+		errCh <- l.Run(ctx)
+	}()
+
+	l.Enqueue(1)
+	l.Close(2)
+
+	// This enqueue should be ignored.
+	l.Enqueue(3)
+
+	wg.Wait()
+
+	got := h.Values()
+	assert.Equal(t, []int{1, 2}, got)
+}
diff --git a/events/loop/segment.go b/events/loop/segment.go
new file mode 100644
index 00000000..011ea99b
--- /dev/null
+++ b/events/loop/segment.go
@@ -0,0 +1,54 @@
+/*
+Copyright 2025 The Dapr Authors
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+    http://www.apache.org/licenses/LICENSE-2.0
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+package loop
+
+// queueSegment is a segment in a linked list of buffered channels.
+// We always write to the tail segment. When the tail segment is full,
+// we create a new tail segment and close the old one so that Run can
+// move on once it has drained it.
+type queueSegment[T any] struct {
+	ch   chan T
+	next *queueSegment[T]
+}
+
+// getSegment gets a queueSegment from the pool or allocates a new one.
+// It always initializes a fresh channel of the configured size.
+func (l *loop[T]) getSegment() *queueSegment[T] {
+	if l.segPool.New == nil {
+		l.segPool.New = func() any {
+			return new(queueSegment[T])
+		}
+	}
+
+	seg := l.segPool.Get().(*queueSegment[T])
+	seg.next = nil
+
+	segSize := l.segSize
+	if segSize == 0 {
+		segSize = 1
+	}
+	seg.ch = make(chan T, segSize)
+
+	return seg
+}
+
+// putSegment returns a segment to the pool after clearing references.
+func (l *loop[T]) putSegment(seg *queueSegment[T]) {
+	if seg == nil {
+		return
+	}
+	seg.ch = nil
+	seg.next = nil
+	l.segPool.Put(seg)
+}