Packet loss happens when data traveling across a network fails to reach its destination. The single biggest cause is network congestion, where routers and switches receive more traffic than their memory buffers can hold, forcing them to discard packets. But congestion is only one of several causes. Hardware problems, wireless interference, misconfigured settings, and even cyberattacks can all trigger packet loss, and the source of the problem determines how noticeable it is and how you fix it.
Network Congestion and Buffer Overflow
Every router and switch along a network path has a limited amount of buffer memory to hold incoming packets while they wait to be forwarded. When traffic volume spikes and packets arrive faster than the device can process them, that buffer fills up. Once it overflows, the router has no choice but to drop incoming packets entirely. This is the most common cause of packet loss on both home and enterprise networks.
Bursty traffic makes this worse. When multiple streams of variable-rate data converge on the same router at the same time, the combined load can overwhelm buffer capacity even if the average traffic level seems manageable. Think of it like a highway on-ramp during rush hour: the road can handle steady traffic just fine, but a sudden wave of cars merging at once creates a bottleneck.
Modern routers use techniques called active queue management to soften the blow. One common method, called Random Early Detection, monitors how full a buffer is getting and starts dropping a small percentage of packets early, before the buffer is completely full. This signals the sending devices to slow down their transmission rate, preventing a total overflow. It trades a small, controlled amount of packet loss for much better overall network stability.
Wireless Interference and Signal Weakness
Wireless networks are significantly more prone to packet loss than wired connections. Radio frequency interference from other Wi-Fi networks, Bluetooth devices, microwaves, and cordless phones can corrupt data in transit. When a packet arrives garbled, the receiving device discards it.
Physical distance from the router matters too. The farther you are, the weaker the signal, and weaker signals produce more transmission errors. Walls, floors, metal objects, and even large appliances between you and the router all degrade signal strength further. If you’re experiencing packet loss on Wi-Fi but not on a wired connection, the wireless link is almost certainly the bottleneck.
Faulty Hardware and Cables
Damaged or deteriorating network equipment is a sneaky cause of packet loss because it can be intermittent and hard to pin down. A network adapter with corrupted or incompatible drivers may silently drop connections. A bent or worn Ethernet cable can introduce enough electrical noise to corrupt packets. Aging routers and switches with failing memory or overheating components may start discarding packets under load even when traffic levels are well within normal range.
This type of packet loss often shows a pattern: it appears consistently at one specific point in the network path rather than fluctuating with overall traffic. If replacing a cable or updating a network adapter’s drivers eliminates the problem, hardware was the culprit.
MTU Mismatches and Fragmentation
Every network link has a maximum transmission unit (MTU), the largest packet size it can carry without breaking the data into smaller pieces. The standard MTU for Ethernet is 1,500 bytes. When a packet larger than the MTU arrives at a router, one of two things happens: the router splits it into smaller fragments, or it drops the packet entirely.
Whether the packet gets fragmented or dropped depends on the protocol. With IPv4, routers will typically split oversized packets on behalf of the sender unless the packet’s “do not fragment” flag is set. IPv6 routers, however, never fragment packets. They simply drop oversized ones and send an error message back to the source, telling it to send smaller packets next time.
Problems get worse when different segments of a network path have different MTU sizes. If one device along the route supports jumbo frames (packets larger than 1,500 bytes) and another doesn’t, the second device will drop every oversized packet it receives. Networks that use tunneling or encapsulation protocols are especially vulnerable, because wrapping a packet in an extra header increases its size. A packet that was exactly 1,500 bytes becomes too large once the tunnel header is added, and the edge device drops it.
DDoS Attacks and Security Threats
Distributed Denial of Service (DDoS) attacks cause packet loss deliberately. Attackers use networks of compromised computers to flood a target with enormous volumes of traffic, overwhelming its ability to process legitimate packets. A “heavy hitter” attack, for example, generates massive traffic flows aimed at exhausting a server’s resources so that real users’ packets get dropped.
ICMP flood attacks work similarly, bombarding a target with ping requests until its network interfaces can’t keep up. The result is the same buffer overflow that causes congestion-related packet loss, just triggered intentionally and at a much larger scale. Mitigation typically involves rate limiting, blocking suspicious traffic flows, or blacklisting attacking IP addresses, but even with defenses in place, some legitimate traffic can be lost during the attack.
How Packet Loss Affects Different Applications
The impact of packet loss depends heavily on which transport protocol an application uses. Applications built on TCP, like web browsing and file downloads, automatically detect and retransmit lost packets. This preserves data integrity (TCP achieves up to 99.9% delivery success) but adds delay: each retransmission cycle costs roughly 100 to 200 milliseconds. Under heavy packet loss, TCP also throttles its sending speed, which can tank throughput for a long time even after the network recovers.
Applications using UDP, like video calls, online gaming, and live streaming, take the opposite approach. UDP doesn’t retransmit lost packets. It prioritizes speed, accepting that 1% to 10% of packets may be lost depending on conditions. A lost packet in a voice call means a brief audio dropout. In a game, it means a character teleporting or a delayed input. Because UDP never slows down to wait for missing data, it keeps latency low (as little as 100 milliseconds), but you feel every lost packet directly.
Thresholds That Actually Matter
Not all packet loss is noticeable. For general web browsing and file transfers, loss rates under 1% to 2% are absorbed by TCP’s retransmission without much visible impact. Real-time applications are far less forgiving. For VoIP calls, less than 1% packet loss is the standard target. Even at 1%, you’ll hear occasional dropouts. At 5%, the call becomes essentially unusable. Online gaming is even more sensitive, with 0.5% or lower recommended for smooth, responsive play.
How to Find Where Packets Are Dropping
The most useful diagnostic tool for tracking down packet loss is MTR, which combines the functions of ping and traceroute into a single test. It sends packets along every hop of the network path between your device and a destination, then reports the loss percentage at each hop. Running a command like mtr -c 100 [destination IP] --report sends 100 test packets and generates a summary showing exactly where in the path packets are disappearing.
The key is reading the results correctly. If one hop in the middle shows high loss but all subsequent hops look fine, that device is likely just deprioritizing test traffic (many routers do this) and it’s not a real problem. If loss appears at a specific hop and persists through every hop after it, that’s where the actual issue lives. You can also run TCP-based MTR tests on a specific port to check whether a firewall or application-level issue is responsible rather than the network itself.