diff --git a/include/zephyr/net/net_if.h b/include/zephyr/net/net_if.h
index 4e8aa607aee9..b3ca7de663d4 100644
--- a/include/zephyr/net/net_if.h
+++ b/include/zephyr/net/net_if.h
@@ -614,7 +614,8 @@ struct net_traffic_class {
 	/** Fifo for handling this Tx or Rx packet */
 	struct k_fifo fifo;
 
-#if NET_TC_COUNT > 1 || defined(CONFIG_NET_TC_SKIP_FOR_HIGH_PRIO)
+#if NET_TC_COUNT > 1 || defined(CONFIG_NET_TC_SKIP_FOR_HIGH_PRIO) \
+	|| defined(CONFIG_NET_TC_RX_SKIP_FOR_HIGH_PRIO)
 	/** Semaphore for tracking the available slots in the fifo */
 	struct k_sem fifo_slot;
 #endif
diff --git a/subsys/net/ip/Kconfig b/subsys/net/ip/Kconfig
index 6853869a3ccb..40c158e1be39 100644
--- a/subsys/net/ip/Kconfig
+++ b/subsys/net/ip/Kconfig
@@ -407,6 +407,19 @@ config NET_TC_SKIP_FOR_HIGH_PRIO
 	  be pushed directly to network driver and will skip the traffic class
 	  queues. This is currently not enabled by default.
 
+config NET_TC_RX_SKIP_FOR_HIGH_PRIO
+	bool "Push high priority packets directly to the application"
+	help
+	  If this is set, then high priority (>= NET_PRIORITY_CA) net_pkt will
+	  be pushed directly to the application and will skip the traffic class
+	  queues. If you select this option, then it means that high priority
+	  network traffic is pushed from the driver to the application thread
+	  without any intermediate RX queue. If the network device driver is
+	  running in IRQ context, it will handle the packet all the way to the
+	  application. This might cause other incoming packets to be lost if
+	  the RX processing takes long time.
+	  This is currently not enabled by default.
+
 choice NET_TC_THREAD_TYPE
 	prompt "How the network RX/TX threads should work"
 	help
diff --git a/subsys/net/ip/net_core.c b/subsys/net/ip/net_core.c
index 6257a2954db6..4c22e10f35a2 100644
--- a/subsys/net/ip/net_core.c
+++ b/subsys/net/ip/net_core.c
@@ -475,7 +475,8 @@ static void net_queue_rx(struct net_if *iface, struct net_pkt *pkt)
 	NET_DBG("TC %d with prio %d pkt %p", tc, prio, pkt);
 #endif
 
-	if (NET_TC_RX_COUNT == 0) {
+	if ((IS_ENABLED(CONFIG_NET_TC_RX_SKIP_FOR_HIGH_PRIO) &&
+	     prio >= NET_PRIORITY_CA) || NET_TC_RX_COUNT == 0) {
 		net_process_rx_packet(pkt);
 	} else {
 		if (net_tc_submit_to_rx_queue(tc, pkt) != NET_OK) {
diff --git a/subsys/net/ip/net_tc.c b/subsys/net/ip/net_tc.c
index 046123aeb2f7..8cb203ad05a7 100644
--- a/subsys/net/ip/net_tc.c
+++ b/subsys/net/ip/net_tc.c
@@ -18,12 +18,15 @@ LOG_MODULE_REGISTER(net_tc, CONFIG_NET_TC_LOG_LEVEL);
 #include "net_stats.h"
 #include "net_tc_mapping.h"
 
-#if NET_TC_RX_COUNT > 1
-#define NET_TC_RX_SLOTS (CONFIG_NET_PKT_RX_COUNT / NET_TC_RX_COUNT)
+#define TC_RX_PSEUDO_QUEUE (COND_CODE_1(CONFIG_NET_TC_RX_SKIP_FOR_HIGH_PRIO, (1), (0)))
+#define NET_TC_RX_EFFECTIVE_COUNT (NET_TC_RX_COUNT + TC_RX_PSEUDO_QUEUE)
+
+#if NET_TC_RX_EFFECTIVE_COUNT > 1
+#define NET_TC_RX_SLOTS (CONFIG_NET_PKT_RX_COUNT / NET_TC_RX_EFFECTIVE_COUNT)
 BUILD_ASSERT(NET_TC_RX_SLOTS > 0,
 		"Misconfiguration: There are more traffic classes then packets, "
 		"either increase CONFIG_NET_PKT_RX_COUNT or decrease "
-		"CONFIG_NET_TC_RX_COUNT");
+		"CONFIG_NET_TC_RX_COUNT or disable CONFIG_NET_TC_RX_SKIP_FOR_HIGH_PRIO");
 #endif
 
 #define TC_TX_PSEUDO_QUEUE (COND_CODE_1(CONFIG_NET_TC_SKIP_FOR_HIGH_PRIO, (1), (0)))
@@ -84,7 +87,7 @@ enum net_verdict net_tc_submit_to_rx_queue(uint8_t tc, struct net_pkt *pkt)
 #if NET_TC_RX_COUNT > 0
 	net_pkt_set_rx_stats_tick(pkt, k_cycle_get_32());
 
-#if NET_TC_RX_COUNT > 1
+#if NET_TC_RX_EFFECTIVE_COUNT > 1
 	if (k_sem_take(&rx_classes[tc].fifo_slot, K_NO_WAIT) != 0) {
 		return NET_DROP;
 	}
@@ -272,7 +275,7 @@ static void tc_rx_handler(void *p1, void *p2, void *p3)
 	ARG_UNUSED(p3);
 
 	struct k_fifo *fifo = p1;
-#if NET_TC_RX_COUNT > 1
+#if NET_TC_RX_EFFECTIVE_COUNT > 1
 	struct k_sem *fifo_slot = p2;
 #else
 	ARG_UNUSED(p2);
@@ -285,7 +288,7 @@ static void tc_rx_handler(void *p1, void *p2, void *p3)
 			continue;
 		}
 
-#if NET_TC_RX_COUNT > 1
+#if NET_TC_RX_EFFECTIVE_COUNT > 1
 		k_sem_give(fifo_slot);
 #endif
 
@@ -429,7 +432,7 @@ void net_tc_rx_init(void)
 
 		k_fifo_init(&rx_classes[i].fifo);
 
-#if NET_TC_RX_COUNT > 1
+#if NET_TC_RX_EFFECTIVE_COUNT > 1
 		k_sem_init(&rx_classes[i].fifo_slot, NET_TC_RX_SLOTS, NET_TC_RX_SLOTS);
 #endif
 
@@ -437,7 +440,7 @@ void net_tc_rx_init(void)
 				      K_KERNEL_STACK_SIZEOF(rx_stack[i]),
 				      tc_rx_handler,
 				      &rx_classes[i].fifo,
-#if NET_TC_RX_COUNT > 1
+#if NET_TC_RX_EFFECTIVE_COUNT > 1
 				      &rx_classes[i].fifo_slot,
 #else
 				      NULL,