Packet Loss by Connection Type Report 2026
By SpeedTestHQ Research · Updated June 16, 2026
Packet loss is the silent killer of real-time apps: a connection can post a great download speed yet still drop the packets that make calls choppy and games rubberband. This report compares how much packet loss each connection type typically shows, what causes it, and where the loss usually happens.
Key findings
- Wired beats wireless for consistency. Fiber and well-provisioned cable normally sit near zero loss; wireless access types vary far more.
- Most home packet loss is local. Wi-Fi interference, weak signal, and an overloaded router cause more loss than the ISP path for the typical household.
- Loss spikes under load. A connection that shows 0% idle can lose packets once the upstream is saturated — the same mechanism behind bufferbloat.
- Small percentages matter. Even 1-2% sustained loss is enough to degrade voice, video calls, and competitive gaming, while web pages may still feel fine.
Typical packet loss by connection type
| Connection type | Typical idle loss | Loss under load / peak | Main causes |
|---|---|---|---|
| Fiber (FTTH) | 0-0.1% | Low | Rare; usually router or Wi-Fi, not the line |
| Cable (DOCSIS) | 0-0.3% | Low to medium | Shared node congestion, upstream saturation, plant noise |
| 5G home internet | 0.2-1.5% | Medium | Tower load, radio interference, signal variability |
| DSL | 0.2-1% | Medium | Line quality, distance from exchange, small upstream |
| Fixed wireless | 0.3-2% | Medium to high | Line-of-sight, weather, sector load, antenna alignment |
| LEO satellite | 0.3-2% | Medium | Beam handoff, obstructions, peak-hour congestion |
| GEO satellite | 1-3% | High | Long path, weather fade, congestion |
| Wi-Fi (any uplink) | 0-3%+ | Highly variable | Interference, weak signal, channel congestion, distance |
What counts as too much
| Sustained loss | Experience | Severity |
|---|---|---|
| 0% | Ideal; nothing to fix | Excellent |
| Under 0.5% | Fine for gaming and calls | Good |
| 0.5-1% | Occasional voice/video artifacts | Acceptable |
| 1-2.5% | Noticeable stutter, dropouts, rubberbanding | Needs attention |
| 2.5-5% | Calls and games frequently unstable | Severe |
| 5%+ | Real-time apps largely unusable | Critical |
Brief bursts of loss matter more for real-time traffic than for downloads. A file transfer simply retransmits the missing packets and finishes a little slower; a voice or game packet that arrives late is useless and gets discarded, which you hear and feel immediately.
Where the loss usually happens
- Inside your home (most common): Wi-Fi interference, distance from the router, and channel congestion. Test over Ethernet to rule this out first.
- The local access link: a noisy cable plant, a weak DSL line, or a marginal wireless signal.
- Under saturation: a full upstream queue drops packets during big uploads, backups, or many devices at once.
- The ISP path or peering: congested nodes or routes, often worst at peak hours.
How to diagnose it
- Test wired first: connect a computer by Ethernet and re-test to separate Wi-Fi loss from line loss.
- Run a sustained test, not a one-second check — loss is intermittent and a quick test can miss it. See how to test for packet loss.
- Compare idle vs loaded: start a large upload and watch whether loss appears only under load.
- Test at multiple times of day; peak-hour-only loss points to access-network congestion.
- For evidence to give your ISP, capture packets for the ticket.
Methodology
This report compares common access technologies using SpeedTestHQ loss scenarios: idle, upload-saturated, and mixed household load, across wired and Wi-Fi paths. The figures are a planning model based on typical consumer deployments and the known behavior of each access type, not a measurement of any single address.
These figures are planning ranges, not a guarantee for every address or device. Your result can change with router placement, local interference, server distance, ISP routing, plan tier, firmware, client hardware, and time of day. For your own connection, run a wired test and compare it with Wi-Fi and peak-hour tests.
Frequently Asked Questions
What is an acceptable amount of packet loss?
Under 1% is generally fine for browsing and streaming. For gaming and video calls, aim for under 0.5%, ideally near 0%. Sustained loss above 2-3% causes noticeable stutter, dropouts, and rubberbanding.
Which connection type has the most packet loss?
Wireless access types — fixed wireless, 5G home internet, and satellite — tend to show more variable packet loss than wired fiber or cable, because radio links are affected by interference, weather, and tower or beam congestion.
Does packet loss happen on my Wi-Fi or the internet?
Both are possible. Wi-Fi interference and weak signal are the most common home cause. Test over Ethernet to separate local Wi-Fi loss from loss on the ISP path before contacting support.
Related Reports and Guides
Bufferbloat by Connection Type
Why latency spikes when the connection is busy.
Gaming Latency Report
Ping by ISP and game, where loss hurts most.
How to Fix Packet Loss
Step-by-step fixes for a lossy connection.
What Is Packet Loss?
A plain-English explanation of the problem.
Ping & Loss Test
Check your own latency and stability.