diff --git a/src/go/rpk/pkg/adminapi/api_partition.go b/src/go/rpk/pkg/adminapi/api_partition.go
index 574c1a1e3072..a182fe7fa300 100644
--- a/src/go/rpk/pkg/adminapi/api_partition.go
+++ b/src/go/rpk/pkg/adminapi/api_partition.go
@@ -204,3 +204,8 @@ func (a *AdminAPI) TransferLeadership(ctx context.Context, fs afero.Fs, p *confi
 	path := fmt.Sprintf("/v1/partitions/%s/%s/%d/transfer_leadership?target=%s", ns, topic, partition, target)
 	return cl.sendOne(ctx, http.MethodPost, path, nil, nil, false)
 }
+
+// TriggerBalancer Trigger on-demand balancer.
+func (a *AdminAPI) TriggerBalancer(ctx context.Context) error {
+	return a.sendToLeader(ctx, http.MethodPost, "/v1/partitions/rebalance", nil, nil)
+}
diff --git a/src/go/rpk/pkg/cli/cluster/partitions/balancer_run.go b/src/go/rpk/pkg/cli/cluster/partitions/balancer_run.go
new file mode 100644
index 000000000000..fff5e9a2ab1c
--- /dev/null
+++ b/src/go/rpk/pkg/cli/cluster/partitions/balancer_run.go
@@ -0,0 +1,67 @@
+// Copyright 2024 Redpanda Data, Inc.
+//
+// Use of this software is governed by the Business Source License
+// included in the file licenses/BSL.md
+//
+// As of the Change Date specified in that file, in accordance with
+// the Business Source License, use of this software will be governed
+// by the Apache License, Version 2.0
+
+package partitions
+
+import (
+	"fmt"
+
+	"github.com/redpanda-data/redpanda/src/go/rpk/pkg/adminapi"
+	"github.com/redpanda-data/redpanda/src/go/rpk/pkg/config"
+	"github.com/redpanda-data/redpanda/src/go/rpk/pkg/out"
+	"github.com/spf13/afero"
+	"github.com/spf13/cobra"
+)
+
+func newTriggerBalancerCommand(fs afero.Fs, p *config.Params) *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "balance",
+		Short: "Triggers on-demand partition balancing",
+		Long: `Triggers on-demand partition balancing.
+
+This command allows you to trigger on-demand partition balancing.
+
+With Redpanda Community Edition, the partition count on each broker
+can easily become uneven, which leads to data skewing. To distribute
+partitions across brokers, you can run this command to trigger
+on-demand partition balancing.
+
+With Redpanda Enterprise Edition, Continuous Data Balancing monitors
+broker and rack availability, as well as disk usage, to avoid topic
+hotspots. However, there are edge cases where users should manually
+trigger partition balancing (such as a node becoming unavailable for
+a prolonged time and rejoining the cluster thereafter). In such cases,
+you should run this command to trigger partition balancing manually.
+
+After you run this command, monitor the balancer progress using:
+
+    rpk cluster partitions balancer-status
+
+To see more detailed movement status, monitor the progress using:
+
+    rpk cluster partitions move-status
+`,
+
+		Args: cobra.ExactArgs(0),
+		Run: func(cmd *cobra.Command, _ []string) {
+			p, err := p.LoadVirtualProfile(fs)
+			out.MaybeDie(err, "rpk unable to load config: %v", err)
+			config.CheckExitCloudAdmin(p)
+
+			cl, err := adminapi.NewClient(fs, p)
+			out.MaybeDie(err, "unable to initialize admin client: %v", err)
+
+			err = cl.TriggerBalancer(cmd.Context())
+			out.MaybeDie(err, "failed to invoke on-demand partition balancing: %v", err)
+
+			fmt.Println("Successfully triggered on-demand partition balancing.\n\nPlease find the progress with 'rpk cluster partitions balancer-status'.")
+		},
+	}
+	return cmd
+}
diff --git a/src/go/rpk/pkg/cli/cluster/partitions/partitions.go b/src/go/rpk/pkg/cli/cluster/partitions/partitions.go
index 1d4333ff6a6b..e6c2925ce190 100644
--- a/src/go/rpk/pkg/cli/cluster/partitions/partitions.go
+++ b/src/go/rpk/pkg/cli/cluster/partitions/partitions.go
@@ -33,6 +33,7 @@ func NewPartitionsCommand(fs afero.Fs, p *config.Params) *cobra.Command {
 		newPartitionEnableCommand(fs, p),
 		newPartitionMovementsStatusCommand(fs, p),
 		newTransferLeaderCommand(fs, p),
+		newTriggerBalancerCommand(fs, p),
 		newUnsafeRecoveryCommand(fs, p),
 	)
 	return cmd