Mbps to MB/s Converter

Mbps to MB/s — quick reference

The formula, worked line by line

Divide by eight. Everything else on this page is about what the answer means, because the arithmetic itself has no options in it: a byte is eight bits, and both units carry the same decimal mega.

Keeping those two claims apart is what makes the number useful. Stated as a ceiling it is reliable and testable. Stated as a prediction it will be wrong most of the time, and the wrongness will be blamed on the arithmetic rather than on the network.

megabytes per second = megabits per second / 8
megabits per second = megabytes per second x 8
100 / 8 = 12.5 MB/s
1,000 / 8 = 125 MB/s
seconds = file size in MB / rate in MB/s

Where the six percent goes

A standard Ethernet frame occupies 1,538 byte-times on the wire for 1,500 bytes of payload: 14 bytes of header, 4 of frame check sequence, 8 of preamble and 12 of interframe gap. IP and TCP headers take a further 40 bytes from inside the payload, and TCP options can take a dozen more, which lands the application-layer share around 94 percent.

The same conversion at other scales

It is a small thing but a genuinely useful one. Any bit rate that is a round power of ten gives a byte rate of 125 in some unit, so a gigabit is 125 megabytes a second, a megabit is 125 kilobytes a second, and a hundred megabits is a tenth of a gigabit and therefore 12.5.

How to use the mbps to MB/s converter

Divide by eight. A connection sold at 100 Mbps carries 12.5 megabytes per second, because a byte is eight bits and both units use the same decimal mega. That is an exact unit conversion, not an estimate, and it accounts for almost all of the gap between what a plan advertises and what a download manager displays.

The distinction worth holding onto is what the answer means. Twelve and a half megabytes per second is the ceiling the line imposes. It is not a promise about any particular transfer, because a transfer also depends on protocol overhead, the server at the other end, the path between you, and whatever else is using the connection at the time.

The reason the two units differ at all is that they come from different industries. Communications engineering counts bits, because a link carries individual signal transitions and a byte is not a meaningful unit on the wire. Storage counts bytes, because that is the smallest independently addressable unit. Your download crosses from one world into the other, and the factor of eight applies at the boundary.

After the factor of eight comes a much smaller correction. Ethernet wraps each frame in 38 bytes of preamble, header, checksum and interframe gap, and IP and TCP take another 40 from inside the payload. On a standard 1,500-byte frame that leaves roughly 94 percent of the line rate available to an application, which is a few percent rather than a factor.

Why a download looks slower than the plan

Three things separate an advertised speed from a number on a progress bar, and they are very different in size. The factor of eight is the largest by an order of magnitude and is pure unit conversion. Protocol overhead is a few percent. Everything else — the server, the path, contention on your own network — is variable and often dominates in practice.

That is the useful diagnostic. Once you have converted properly, a transfer that lands near the calculated figure tells you the connection is working as sold. A transfer far below it points somewhere else entirely, and no further arithmetic about units will explain it.

What a speed test is and is not measuring

A speed test opens several parallel connections to a nearby, well-provisioned server and measures a short burst. That is deliberately favourable, and it is the right design for its purpose: it tells you what your line can do when nothing else is in the way. It does not tell you what any particular download will do.

The other half of the picture is upload, which on many connections is far lower than download and is what determines how quickly you can send a large file or hold a video call. The same division by eight applies, and the same distinction between a line rate and what an application sees.

Every Mbps value, worked out

214 common Mbps figures each get their own page, with the answer at full precision, the arithmetic, what rounding costs, and the nearest real-world reference point on the scale.

Questions people ask