Convert kB to MB

A kilobyte is 1,000 bytes when a standards body defines it and 1,024 when an operating system does. This page uses 1,000; the kibibyte is the unit that means 1,024.

A megabyte is a million bytes. Storage manufacturers have always used this decimal meaning, which is why their capacities look larger than what a computer reports.

Going from kilobytes to megabytes moves the decimal point 3 places to the left and changes nothing else. There is no factor to remember and no rounding to decide: the digits stay in the same order and only their position changes.

1,234 kB is 1.234 MB — same digits, moved along. That is worth knowing because it is the one kind of conversion you can check at a glance: if the digits of the answer are not the digits you started with, something other than the conversion has happened to them.

The megabyte on this page is a million bytes, which makes it a thousand kilobytes and the conversion three digits moved off the right. The prefixes behave the way they do everywhere else in measurement: kilo is a thousand, mega is a million, and each rung of the ladder is the same factor above the last. That consistency is the practical argument for the decimal system, because it lets a size be moved up or down the ladder in the head.

The binary pair — kibibyte and mebibyte — has 1,024 in both places, so a mebibyte is 1,048,576 bytes and sits 4.9% above a megabyte. Two rungs is enough for the gap to be visible: a total of 5,000 kB is 5 MB decimal but only 4.77 in the binary units. Most of the time the difference is smaller than the error in the total itself, and the exception is a limit, where 5% is the whole margin.

Almost nobody converts one kilobyte figure to megabytes; they convert a sum. The list comes from a build report, a mail client, a file dialogue, an export. The safe procedure is to add in kilobytes and convert once, because converting each row first and rounding it introduces an error per row that accumulates in one direction — sizes quoted publicly are rounded down more often than up.

The second thing a list hides is what it excludes. A build report usually counts the assets it produced and not the fonts or images referenced from a stylesheet; a mail client counts attachments and not the message body or the inline images; a file dialogue counts files and not the directories or the space they occupy. The megabyte total is exactly as complete as the list, and the list is where a surprise comes from, not the division.

The high-density 3.5-inch disk held 1,474,560 bytes. That is 1,440 kibibytes, and the number on the label was produced by dividing by 1,024 once and then by 1,000 again — so the megabyte in "1.44 MB" was 1,024,000 bytes, a hybrid that exists in no standard and was invented by the marketing of a single product. In the units on this page the disk held 1.47 MB; in binary units it held 1.41 MiB.

It is the clearest illustration in the whole category of what goes wrong when a label is chosen after the arithmetic rather than before it. Nobody was lying and every figure on the box was derived from a real byte count, and the result was still a unit that means nothing. The same instinct — divide by whichever number makes the number look right, then print a familiar suffix — is what produces disagreements between file managers today.

A Windows file listing showing "1,024 KB" has divided a byte count by 1,024, so its kilobytes are kibibytes under another name. Adding those figures and dividing by 1,000 mixes the systems: the result is neither a clean megabyte nor a clean mebibyte, and it will be about 2.4% away from whichever one the reader assumes. Over a list of a hundred files that is a consistent bias rather than noise.

The way out is to take the byte count where one is available. Windows shows it in the properties dialogue and every scripting tool exposes it, and the byte count is the number both systems agree on. If only the rounded kilobyte column exists, the total is fine for a rough judgement and should not be used against a limit that the file will be measured against by something else.

A web page arrives as dozens of separate assets, each of a size that looks harmless: a font at 24 kB, a script at 140, a hero image at 380, an analytics bundle at 90. None of them is worth an argument alone, and the sum crosses a megabyte without any single decision having been made to spend it. Converting the list is how the sum becomes visible.

The figure to convert is the transferred size rather than the raw one, because text assets are compressed on the way and images are not compressed again. A build report that lists raw kilobytes will overstate the scripts by roughly a factor of three and be right about the images, which is the worst of both — the total is wrong and it is wrong unevenly, so the largest-looking item is not the one worth removing.

One decimal place on a megabyte figure carries a hundred kilobytes and two carry ten, so the right number of digits depends entirely on what the total is for. A page-weight budget is argued about in tenths, because a tenth of a megabyte is a whole font. A photo library is not, because a hundred kilobytes either way is not a decision anybody is going to take.

What is never worth doing is quoting more decimals than the input had. A list of sizes rounded to whole kilobytes cannot support three decimal places in megabytes, however confidently a spreadsheet prints them. If the underlying byte counts are available the extra digits are real; if the list was already rounded, the digits are the rounding of the rounding.

Kilobytes stay readable to about four digits. Past ten thousand, the figures stop being comparable at a glance — 9,400 kB and 14,200 kB take a moment to place, while 9.4 MB and 14.2 MB do not — and that is the practical signal to convert. It is the same reason a document does not quote a person’s height in millimetres despite the number being perfectly correct.

The reverse also applies. A total of 0.06 MB is worse than 60 kB, and a list where the smallest item rounds to 0.0 MB has lost the information it was meant to carry. Pick the unit that leaves the interesting part of the number to the left of the decimal point, and quote the byte count instead of either when the figure is going to be checked by a machine.