How to Estimate Download Time Without Getting Burned by Bits vs Bytes

The first time I tried to predict how long a download would take, I was wrong by a factor of eight. I had a 1 GB file and a "100 Mbps" connection, did the obvious arithmetic in my head, and confidently said "about ten seconds." Eighty seconds later the file finally landed, and I went looking for what I had missed. The answer was a single hidden conversion that trips up almost everyone who tries to do this math by hand.

If you only remember one thing from this post, make it this: your internet speed is measured in bits, your file is measured in bytes, and there are 8 bits in a byte. Skip that step and every estimate you make will be eight times too optimistic.

The one equation that actually matters

The core relationship is simple once the units agree:

time = file size in bits ÷ speed in bits per second

The trap is that the two inputs come in different units by default. File sizes are quoted in bytes (KB, MB, GB), while connection speeds are sold in bits per second (Kbps, Mbps, Gbps). So before you divide, you have to put both numbers into the same currency.

The cleanest way to do that is to convert your speed into bytes per second and divide the file size by it. The conversion is just division by 8:

100 Mbps ÷ 8 = 12.5 MB/s

That 12.5 MB/s is the number your download dialog would show if it labelled things honestly. A "100 Mbps" plan does not move 100 megabytes per second — it moves about 12.5. Once you internalize that, most of the confusion evaporates.

A worked example: a 5 GB file on a 100 Mbps line

Let me walk through a real case end to end so the arithmetic is concrete.

You want to download a 5 GB file on a 100 Mbps connection. Here is the whole chain:

Convert the speed to bytes per second: 100 Mbps ÷ 8 = 12.5 MB/s.
Divide the file by the throughput: 5000 MB ÷ 12.5 MB/s = 400 seconds.
Convert to something human: 400 ÷ 60 ≈ 6.7 minutes.

So a 5 GB file on a 100 Mbps line takes about 6.7 minutes under ideal conditions. Notice how the units stayed consistent the whole way: I expressed the 5 GB as 5000 MB (decimal, 1 GB = 1000 MB) so it matched the megabytes-per-second figure I got from the speed conversion. Mixing decimal sizes with anything else is where the next batch of mistakes comes from.

If you would rather not run this by hand every time, the Download Time Calculator does the bits-to-bytes step for you and prints the effective MB/s right next to the time, so the two numbers never end up fighting each other.

Decimal vs binary: the second unit trap

There is a quieter version of the same problem hiding in the word "gigabyte." Two different definitions are in circulation:

Decimal: 1 GB = 1000 MB = 1,000,000,000 bytes. This is what your internet contract and the box your hard drive came in use.
Binary: 1 GiB = 1024 MiB. This is what most operating systems report when they show file sizes, even though they often still write "GB."

The gap between them is about 7% at the gigabyte scale, and it grows as the numbers get bigger. For transfer-time math it is cleanest to keep everything decimal, because that matches the advertised speed your ISP quotes. If you grab a size your OS reported in binary and divide it by a decimal speed, your answer drifts by a few percent — small, but enough to be annoying when you are timing a deadline. If you want to nail down exactly how many bytes a "240 GB" drive really holds in each base, the Data Storage Converter will translate between GB, GiB, TB and TiB cleanly.

Why the real download is always slower than the math

Here is the part the textbook equation never mentions: the number you calculate is a theoretical floor, not a promise. It assumes the full advertised speed reaches your file with zero friction, and reality is full of friction. The usual suspects:

Protocol overhead. TCP, TLS and HTTP headers eat roughly 5 to 15 percent of your raw bandwidth before any actual file data moves. That alone turns a clean 6.7-minute estimate into something closer to 7.5.
The server on the other end. A distant, busy, or rate-limited server can cap your speed far below your local link. Your gigabit fiber does nothing if the download host only feeds you 20 Mbps.
Wi-Fi and shared lines. Walls, interference, and a roommate streaming 4K all carve into your effective throughput. A wired connection often beats Wi-Fi by a wide margin here.
Peak vs sustained speed. ISPs advertise the best case. The speed you actually hold for a full ten-minute download is usually lower than the burst you see in the first few seconds.

My rule of thumb: take the theoretical time, add 15 to 25 percent, and quote that if anyone is depending on the answer. It is the difference between "the calculator said six minutes" and "it'll be done in about eight, give or take."

Working backwards from a deadline

Sometimes the question is reversed. You do not have a speed and want a time — you have a deadline and need to know whether your connection can even hit it. The same equation rearranges:

required speed in bits/sec = file size in bits ÷ target seconds

Say a client needs a 5 GB video master delivered in 5 minutes (300 seconds). Convert the file to bits (5 GB × 8 = 40 billion bits), divide by 300, and you get about 133 Mbps of sustained upload required. If your plan only uploads at 30 Mbps, that deadline is physically impossible over that link — better to ship a drive than to watch a progress bar miss the mark.

And do not forget that home connections are usually asymmetric. A plan sold as "300 Mbps" almost always means download; the upload might be a tenth of that. A 2 GB video that downloads in under a minute can take eight minutes or more to push back up.

The short version

Speed is in bits, files are in bytes, and 8 bits make 1 byte — divide Mbps by 8 to get MB/s.
time = file size in bits ÷ speed in bits per second. A 5 GB file on 100 Mbps takes about 6.7 minutes.
Keep both numbers in the same base (decimal: 1 GB = 1000 MB) so they line up with your contract.
Treat the result as a best case and add 15 to 25 percent for overhead, slow servers, and Wi-Fi.

Do the conversion once and the rest is just division. Get the units right, pad for reality, and your estimate will stop embarrassing you.

Made by Toolora · Updated 2026-06-13