LAN Speed Test — How to Measure Your Real File Transfer Speed

Why Measuring Matters

A network card that reports a 1 Gbps link speed does not guarantee 1 Gbps throughput. Real transfer speed is affected by cable quality, switch capacity, drive read/write speed, CPU overhead from encryption, and protocol efficiency. The only way to know what your network actually delivers is to measure it.

This guide covers three methods — from quickest to most accurate — for measuring LAN file transfer speed on Windows and Mac.

Method 1 — Task Manager During a Real Transfer (Windows)

The fastest way to get a real-world number is to watch Task Manager during an actual file transfer.

1. Start copying a large file (at least 1GB, ideally 5GB+) from one machine to another on your LAN — use File Explorer, a mapped drive, or any LAN tool
2. On either machine: press Ctrl + Shift + Esc → Performance tab → select Ethernet or WiFi
3. Watch the "Send" and "Receive" values while the transfer is in progress

Read the throughput in Mbps shown on the graph, or calculate: transfer speed shown in MB/s × 8 = Mbps.

What you are looking for:

• Wired Gigabit: 800–950 Mbps (~100–115 MB/s)
• WiFi 5 (good conditions): 150–400 Mbps
• If consistently below 200 Mbps on wired gigabit: there is a bottleneck

Limitation of this method: You are measuring the combined throughput of network + drive. If the drive is the bottleneck, the network measurement will be artificially low.

Method 2 — Activity Monitor During a Transfer (Mac)

1. Start a large file transfer over the network (Finder → Connect to Server, or AirDrop, or LocalSend)
2. Open Activity Monitor: Spotlight (Cmd+Space) → type Activity Monitor → Network tab
3. Watch "Data received/sec" or "Data sent/sec" depending on whether you are receiving or sending

Alternatively: System Settings → Network → [your connection] shows the current link speed (confirming 1 Gbps link, not actual throughput).

For actual throughput on Mac, the Activity Monitor method during a large transfer is the clearest real-world measurement.

Method 3 — iperf3 (Network-Only Measurement, Windows + Mac)

iperf3 is a free, open-source network benchmarking tool that measures pure network throughput without drive I/O as a variable. It sends test data directly between two machines in memory, isolating the network performance.

Install iperf3:

• Windows: download from iperf.fr → no installation needed, run from a folder
• Mac: install via Homebrew: brew install iperf3
• Linux: sudo apt install iperf3

Run the test: On the machine that will receive (the server):

iperf3 -s

On the machine that will send (the client):

iperf3 -c [SERVER-IP-ADDRESS]

Replace [SERVER-IP-ADDRESS] with the LAN IP of the server machine (find it in Windows: Settings → Network → your connection → shows IPv4 address. On Mac: System Settings → Network → Details → IP Address).

Read the result: iperf3 reports throughput in Mbits/sec at the end of the test. Divide by 8 to get MB/s.

Expected results:

• Wired Gigabit: 920–960 Mbits/sec
• WiFi 5 (close to AP): 150–400 Mbits/sec
• WiFi 6: 300–700 Mbits/sec

If your wired result is significantly below 900 Mbits/sec: Check cable, switch port, and NIC link speed.

Method 4 — Windows Built-In Network Test

For a quick sanity check without installing anything:

Open PowerShell as Administrator and run:

Test-NetConnection -ComputerName [HOSTNAME-OR-IP] -Port 445

This confirms the TCP connection to the SMB port is working. It does not measure throughput but confirms reachability and that SMB is not blocked by firewall.

Method 5 — LAN Speed Test Tool (Windows GUI)

LAN Speed Test (Totusoft) is a free GUI application for Windows that measures LAN throughput by writing and reading a test file across the network. It combines network and storage measurements and gives a clear MB/s result in a simple interface.

1. Download from totusoft.com
2. Run the application
3. Set the test file path to a mapped network drive or a UNC path (e.g., \\SERVERNAME\Share\)
4. Set test file size to 1GB or larger
5. Click Start

The result shows write speed (your machine writing to the remote share) and read speed (reading back). This reflects real-world file transfer performance including drive I/O.

Interpreting Your Results

Measured speed	Likely cause	Action
90–115 MB/s (wired)	Healthy gigabit LAN	No action needed
40–80 MB/s (wired)	Bad cable, wrong MTU, 100Mbps link	Check cable, confirm gigabit link
10–12 MB/s (wired)	Fast Ethernet (100 Mbps) link negotiated	Replace cable, check switch port
30–80 MB/s (WiFi)	Normal WiFi 5 conditions	Acceptable; wire for critical transfers
Under 10 MB/s (wired)	SMB encryption overhead, drive bottleneck, antivirus	Check each variable

Tools That Help With LAN Speed

For daily large file transfers, application-layer tools like Oxolan and LocalSend provide their own progress indicators showing current transfer speed in MB/s — giving you a real-world speed measurement as part of normal use.

Get Oxolan for Windows

Frequently Asked Questions

My iperf3 result is 940 Mbps but actual file transfers are only 70 MB/s. Why? iperf3 measures pure network capacity without drive involvement. If actual transfers are much slower, the bottleneck is drive I/O (reading from source or writing to destination), not the network. Test with CrystalDiskMark to measure drive speed independently.

Can I test WiFi speed between two machines without a server machine? Most practical LAN speed tests need two machines. iperf3 requires one machine as server and one as client. On a single machine, you can test internet speed to a cloud service, but that measures your ISP connection, not your LAN.

Why does my NAS read faster than it writes? HDD-based NAS units typically write slower than they read, especially in RAID 5 configurations (write penalty from parity calculation). RAID 1 (mirroring) has a smaller write penalty. SSD-based NAS units have more symmetrical read/write performance.

Does the iperf3 test use real data or random data? iperf3 generates pseudo-random data for the test. The data is not copied from your files — it is generated in memory on the client and discarded on the server. No files are created on either machine (when run in default mode).