Mbps vs. MB/s: Convert the Units, Not the Guarantee

Eight bits make one byte. Keep decimal and binary prefixes, application overhead, and transfer conditions separate.

By Younes Kirat : Publisher and maintainer
Source-linked guidance ยท Not a comparative benchmark

A megabit is not a megabyte

Mbps means megabits per second. MB/s means megabytes per second. With decimal prefixes, one megabyte is eight megabits: divide Mbps by eight to express the same rate in MB/s, or multiply MB/s by eight to express it in Mbps. Uppercase B denotes bytes; lowercase b denotes bits.

Check the arithmetic

200 Mbps is 25 MB/s as a unit conversion. 20 MB/s is 160 Mbps. The earlier statement that 200 Mbps and 20 MB/s were the same was incorrect and is withdrawn. These examples express units only; they are not measured results or promises of application throughput.

Distinguish decimal and binary units

A decimal megabyte is 1,000,000 bytes. A mebibyte, MiB, is 1,048,576 bytes. An application showing MiB/s needs that additional prefix conversion before comparison with Mbps. Some applications label units ambiguously; check their documentation rather than assuming what a displayed MB means.

Do not infer an exact transfer rate

Protocol framing, encryption, competing traffic, remote-service limits, storage, processing, and application behavior can affect observed transfers. There is no universal overhead percentage that proves a connection healthy. A speed-test throughput estimate and a file application may use different accounting and intervals.

An ideal time calculation is only an illustration

For decimal units, ideal seconds equals file bytes multiplied by eight, divided by bits per second. A 1 GB file at a constant 100 Mbps would take 80 seconds under that simplified model. It excludes overhead, startup, stalls, and other limits, so it is not a predicted download time. Record the units and actual task instead of calling the difference a fault.

Method and sources

Read the HTTP timing limits and validity warnings and testing policy. VelocityVerify does not measure end-to-end packet loss or a separate upload-loaded latency series. New letter grades are withheld.

Editorial update: Earlier calculation or capability claims were revised. See editorial corrections.