Download time, with the units kept honest
Download time is file size divided by speed, but the units hide two traps: connection speeds are sold in bits while files are measured in bytes, and a gigabyte is not a gibibyte. This calculator keeps both separations explicit. Enter the file size in the unit it was actually quoted in, your connection rate in its unit, and an efficiency figure for the overhead real connections carry, and it returns the duration, the effective rate in both Mbps and MB/s, and, if you give a start time, the clock time the download finishes. A measured-speed mode below works out your real rate from a download already under way and projects its remaining time.
The file and the connection
Measured-speed mode (optional)
Disclaimer: This calculator gives estimates for general information only. It is not financial, medical or other professional advice. Real transfer rates vary minute to minute with congestion, server limits and your local network; treat the result as a planning figure.
What this calculator does
It converts a file size and a transfer rate into a duration with the unit discipline the question deserves. File size is converted to bits through the unit you select, with decimal units (powers of 1,000, as drives and download sites quote them) and binary units (powers of 1,024, as operating systems often report them) kept as separate, labelled choices rather than quietly merged. The connection rate, in bits or bytes per second as labelled, is reduced by your efficiency percentage to an effective rate, which is also restated in both Mbps and MB/s so it can be compared with what a download manager is actually showing. The duration is the size in bits over the effective rate, formatted into hours, minutes and seconds, and with a start time it becomes a finish clock time. The measured mode replaces every assumption with observation: it derives your true average rate from data that has already arrived and projects the remainder at that rate.
How to use it
- Match the unit to the quote, not to your preference. A game listed as 85 GB by its store is 85 decimal gigabytes; your operating system will show the same file as about 79 GiB. Entering the store’s 85 against GiB inflates the estimate by 7%; entering the system’s figure against GB deflates it. The unit select exists because both numbers describe the same file and only one pairing is right.
- Divide advertised speed by eight before believing it. A 100 Mbps plan moves at most 12.5 MB/s, and the MB/s figure your browser shows will sit a little below that after overhead. The efficiency field formalises the gap: 90% turns 100 Mbps into an effective 90 Mbps, 11.25 MB/s, which is close to what a healthy connection delivers.
- The server is half the connection. Your rate is the minimum of your line, the server’s per-connection limit, and everything in between. A slow estimate for one download says nothing about the line; the measured mode, run against that very download, measures the combination that actually applies to it.
- Use measured mode for the honest answer mid-download. Progress bars lie early (they average over connection setup and ramp-up). A few minutes in, amount divided by elapsed time is your real rate, and the remaining projection from it beats the plan’s headline figure every time, because it is the same server, the same route and the same evening.
- Uploads use the same page. Enter your upload rate instead of the download rate; the arithmetic is identical and the asymmetry of most plans (fast down, slow up) is exactly why a cloud backup takes all night while the restore took an hour.
How it is calculated
- Size in bits = file size x unit bytes x 8. Headline rate in bits = rate x unit bits. Effective rate = headline rate x efficiency / 100. Time = size in bits / effective rate, in seconds, rounded to the nearest second and formatted as hours, minutes and seconds. Finish time = start time + duration, flagged when it falls after midnight.
- Measured mode: measured rate = amount downloaded / elapsed seconds; remaining time = (file size – amount downloaded) / measured rate, all in consistent byte units selected on the form.
Worked examples
A 4.7 GB file on a 100 Mbps plan at 90% efficiency. The file is 4,700,000,000 bytes, 37,600,000,000 bits. The effective rate is 90 Mbps, which is 11.25 MB/s. Time = 37,600,000,000 / 90,000,000 = 417.78 seconds, shown as 6 minutes 58 seconds. At an ideal 100% efficiency it would be 6 minutes 16 seconds; the overhead costs 42 seconds on this file, and proportionally more on larger ones, which is why the field exists.
10 GiB at 25 MB/s. A binary size this time: 10 GiB is 10,737,418,240 bytes. At 25,000,000 bytes a second the transfer takes 429.50 seconds, shown as 7 minutes 9 seconds. Entered carelessly as 10 GB the estimate would read 6 minutes 40 seconds, an 11% understatement created purely by the unit choice, on identical hardware.
Measured mode on a download in progress. 2 GB has arrived in 5 minutes: a measured 6.67 MB/s, 53.33 Mbps, roughly half the plan’s headline and entirely typical of a single server on a busy evening. The remaining 2.7 GB of the 4.7 GB file projects to 6 minutes 45 seconds more, a finish the browser’s own early estimate, still dreaming of the headline rate, will not admit for several more minutes.
Reading your results
Compare the effective rate with what your download manager shows. If the real figure is far below your efficiency-adjusted estimate, the constraint is not your line but the server, the route, or local Wi-Fi, and changing plans will not fix it; if it matches, the estimate has correctly priced your connection and can be trusted for the next file. Read the duration together with the finish time when you gave one: the practical question is rarely how long but whether it lands before you need the machine, the meeting, or the sleep, and the clock answer is the one you will act on.
Limitations
- Rates are treated as constant. Real transfers ramp, dip and sometimes stall; a single effective rate is a planning model. Measured mode partially cures this by averaging your actual experience, but it still projects that average forward over a connection that may speed up or slow down.
- Per-connection limits are invisible to the headline mode. Many servers cap each connection well below typical line speeds, and download managers open several connections to compensate. If a single stream is capped, no efficiency figure derived from your plan will describe it; measured mode on that download will.
- The efficiency percentage is a judgement, not a constant. Protocol overhead itself is small on modern connections; most of the real-world gap is congestion, Wi-Fi and server behaviour, all of which vary by hour. The 90% suggestion is a planning convention, stated as one, not a property of your line.
- Finish times use your device’s clock. The start time is taken as local time today; a download started before midnight and finishing after is flagged as landing the next day, but multi-day transfers are expressed in the duration rather than as calendar dates.
Frequently asked questions
Why is my download speed in MB/s so much lower than my plan’s Mbps?
Because they are different units measuring the same flow. Plans are sold in megabits per second; software reports megabytes per second; and there are eight bits in a byte. A 100 Mbps plan therefore delivers at most 12.5 MB/s, and with realistic overhead your software will show about 11 MB/s on a healthy connection. Nothing is wrong, nobody is throttling you, and the confusion is the entire reason this page restates the effective rate in both units.
What is the difference between GB and GiB?
A gigabyte (GB) is 1,000,000,000 bytes, the decimal unit drive makers and download stores use. A gibibyte (GiB) is 1,073,741,824 bytes, the binary unit, 2 to the power 30, that operating systems often use when they label a figure “GB”. The same file is a larger number of GB than of GiB, and mixing the figure from one system with the unit of the other misstates a download by about 7% at this scale, growing to about 10% at terabytes. This page offers both, labelled, so the pairing can be made correctly.
What efficiency should I enter?
On a wired connection to a nearby, unloaded server, 95 to 100% is realistic. For a typical home broadband evening over Wi-Fi, 85 to 92% plans honestly. If a specific download matters (a work file before a deadline), skip the guess: start it, let it run two or three minutes, and use measured mode, which replaces the assumed efficiency with your observed rate altogether.
Why did my download finish at a different time than estimated?
Because the estimate froze one rate and the real connection did not. Congestion rises and falls, servers shed load or hit their own limits, and other devices on the network wake and sleep. The estimate’s value is at planning time (will this fit before the meeting, is it worth starting tonight), and measured mode, refreshed as the download proceeds, converges on the true finish as your information improves, which is how estimates are honestly meant to be used.
Related calculators
Method: exact bit and byte conversion with decimal and binary units kept distinct, effective rate after the entered efficiency, duration as size over rate, measured mode from observed progress, exactly as defined on this page. Last reviewed: 13 October 2026.