diff --git a/itest/rcmgr_test.go b/itest/rcmgr_test.go
new file mode 100644
index 0000000000..ce3215d84c
--- /dev/null
+++ b/itest/rcmgr_test.go
@@ -0,0 +1,182 @@
+package itest
+
+import (
+	"context"
+	"sync"
+	"testing"
+	"time"
+
+	rcmgr "github.com/libp2p/go-libp2p-resource-manager"
+
+	"github.com/libp2p/go-libp2p"
+	"github.com/libp2p/go-libp2p-core/peer"
+)
+
+func TestResourceManagerConnInbound(t *testing.T) {
+	// this test checks that we can not exceed the inbound conn limit at system level
+	// we specify: 1 conn per peer, 3 conns total, and we try to create 4 conns
+	limiter := rcmgr.NewFixedLimiter(1 << 30)
+	limiter.SystemLimits = limiter.SystemLimits.
+		WithConnLimit(3, 1024)
+	limiter.DefaultPeerLimits = limiter.DefaultPeerLimits.
+		WithConnLimit(1, 16)
+	echos := createEchos(t, 5, libp2p.ResourceManager(rcmgr.NewResourceManager(limiter)))
+
+	for i := 1; i < 4; i++ {
+		err := echos[i].Host.Connect(context.Background(), peer.AddrInfo{ID: echos[0].Host.ID()})
+		if err != nil {
+			t.Fatal(err)
+		}
+		time.Sleep(10 * time.Millisecond)
+	}
+
+	for i := 1; i < 4; i++ {
+		count := len(echos[i].Host.Network().ConnsToPeer(echos[0].Host.ID()))
+		if count != 1 {
+			t.Fatalf("expected %d connections to peer, got %d", 1, count)
+		}
+	}
+
+	err := echos[4].Host.Connect(context.Background(), peer.AddrInfo{ID: echos[0].Host.ID()})
+	if err == nil {
+		t.Fatal("expected ResourceManager to block incoming connection")
+	}
+}
+
+func TestResourceManagerConnOutbound(t *testing.T) {
+	// this test checks that we can not exceed the inbound conn limit at system level
+	// we specify: 1 conn per peer, 3 conns total, and we try to create 4 conns
+	limiter := rcmgr.NewFixedLimiter(1 << 30)
+	limiter.SystemLimits = limiter.SystemLimits.
+		WithConnLimit(1024, 3)
+	limiter.DefaultPeerLimits = limiter.DefaultPeerLimits.
+		WithConnLimit(16, 1)
+	echos := createEchos(t, 5, libp2p.ResourceManager(rcmgr.NewResourceManager(limiter)))
+
+	for i := 1; i < 4; i++ {
+		err := echos[0].Host.Connect(context.Background(), peer.AddrInfo{ID: echos[i].Host.ID()})
+		if err != nil {
+			t.Fatal(err)
+		}
+		time.Sleep(10 * time.Millisecond)
+	}
+
+	for i := 1; i < 4; i++ {
+		count := len(echos[i].Host.Network().ConnsToPeer(echos[0].Host.ID()))
+		if count != 1 {
+			t.Fatalf("expected %d connections to peer, got %d", 1, count)
+		}
+	}
+
+	err := echos[0].Host.Connect(context.Background(), peer.AddrInfo{ID: echos[4].Host.ID()})
+	if err == nil {
+		t.Fatal("expected ResourceManager to block incoming connection")
+	}
+}
+
+func TestResourceManagerServiceInbound(t *testing.T) {
+	// this test checks that we can not exceed the inbound stream limit at service level
+	// we specify: 3 streams for the service, and we try to create 4 streams
+	limiter := rcmgr.NewFixedLimiter(1 << 30)
+	limiter.DefaultServiceLimits = limiter.DefaultServiceLimits.
+		WithStreamLimit(3, 1024)
+	echos := createEchos(t, 5, libp2p.ResourceManager(rcmgr.NewResourceManager(limiter)))
+
+	echos[0].WaitBeforeRead = func() error {
+		time.Sleep(100 * time.Millisecond)
+		return nil
+	}
+
+	for i := 1; i < 5; i++ {
+		err := echos[i].Host.Connect(context.Background(), peer.AddrInfo{ID: echos[0].Host.ID()})
+		if err != nil {
+			t.Fatal(err)
+		}
+		time.Sleep(10 * time.Millisecond)
+	}
+
+	var wg sync.WaitGroup
+	for i := 1; i < 5; i++ {
+		wg.Add(1)
+		go func(i int) {
+			defer wg.Done()
+
+			err := echos[i].Echo(echos[0].Host.ID(), "hello libp2p")
+			if err != nil {
+				t.Log(err)
+			}
+		}(i)
+	}
+	wg.Wait()
+
+	checkEchoStatus(t, echos[0], EchoStatus{
+		StreamsIn:             4,
+		EchosIn:               3,
+		EchosOut:              3,
+		ResourceServiceErrors: 1,
+	})
+}
+
+func TestResourceManagerServicePeerInbound(t *testing.T) {
+	// this test checks that we cannot exceed the per peer inbound stream limit at service level
+	// we specify: 2 streams per peer for echo, and we try to create 3 streams
+	limiter := rcmgr.NewFixedLimiter(1 << 30)
+	limiter.ServicePeerLimits = map[string]rcmgr.Limit{
+		EchoService: limiter.DefaultPeerLimits.WithStreamLimit(2, 1024),
+	}
+	echos := createEchos(t, 5, libp2p.ResourceManager(rcmgr.NewResourceManager(limiter)))
+
+	echos[0].WaitBeforeRead = func() error {
+		time.Sleep(100 * time.Millisecond)
+		return nil
+	}
+
+	for i := 1; i < 5; i++ {
+		err := echos[i].Host.Connect(context.Background(), peer.AddrInfo{ID: echos[0].Host.ID()})
+		if err != nil {
+			t.Fatal(err)
+		}
+		time.Sleep(10 * time.Millisecond)
+	}
+
+	var wg sync.WaitGroup
+	for i := 1; i < 5; i++ {
+		wg.Add(1)
+		go func(i int) {
+			defer wg.Done()
+
+			err := echos[i].Echo(echos[0].Host.ID(), "hello libp2p")
+			if err != nil {
+				t.Log(err)
+			}
+		}(i)
+	}
+	wg.Wait()
+
+	checkEchoStatus(t, echos[0], EchoStatus{
+		StreamsIn:             4,
+		EchosIn:               4,
+		EchosOut:              4,
+		ResourceServiceErrors: 0,
+	})
+
+	for i := 0; i < 3; i++ {
+		wg.Add(1)
+		go func() {
+			defer wg.Done()
+
+			err := echos[2].Echo(echos[0].Host.ID(), "hello libp2p")
+			if err != nil {
+				t.Log(err)
+			}
+		}()
+	}
+	wg.Wait()
+
+	checkEchoStatus(t, echos[0], EchoStatus{
+		StreamsIn:             7,
+		EchosIn:               6,
+		EchosOut:              6,
+		ResourceServiceErrors: 1,
+	})
+}