The Core Answer: How to Calculate File Size in MB
If you need the short version: to calculate file size in MB, first compute total bits from your media parameters, divide by 8 for bytes, then divide by 1,048,576 for binary MB or 1,000,000 for decimal MB. The universal framework I rely on after years of production work is total bits = dimensions × bit depth × time, bytes = bits ÷ 8, and MB = bytes ÷ (1024² or 1000²). This single method replaces the scattered formulas competitors publish.
When a colleague asks, ‘How do I determine the size of a file?’ I tell them there are two complementary answers. The operating system file properties panel shows the actual stored size, including compression and filesystem overhead. The calculation method reveals the theoretical uncompressed size. You need both to plan storage or debug a mismatch.
I still remember a 2016 client deliverable where I priced a 500-hour audio transcription job based on raw WAV math: 500 × 3600 × 44100 × 16 × 2 ÷ 8 ≈ 317 GB. The supplied MP3s were 14 GB. My quote was absurd because I ignored codec reality. That embarrassment birthed this unified approach.
The Universal Formula and the Variables Behind It
Most articles hand you a different equation for images, sound, and video. In practice, they are the same linear model: information quantity times precision. Once you map each medium’s ‘dimensions’ correctly, the rest is arithmetic.
What Counts as Dimensions, Bit Depth, and Time
For a still image, dimensions are pixel width × pixel height. Bit depth is bits per pixel (e.g., 24-bit RGB). Time equals 1. For audio, dimensions become sample rate (Hz) × duration (seconds) × channels. Bit depth is bits per sample (16-bit CD quality). For video, you can treat dimensions as frame pixel count × frames, where frames = duration × frame rate, and bit depth is bits per pixel per channel, but in real workflows we shortcut to bitrate × time because codecs interleave data.
For text, dimensions are character count; bit depth is 8 bits for ASCII or 16 for UTF-16. Time is again 1. The product of these gives raw bits. Divide by 8 to get bytes. This is the uncompressed footprint before any container, metadata, or compression.
Binary vs Decimal: The MB Confusion
The thing nobody tells you about: ‘MB’ is ambiguous. Drive manufacturers use decimal megabytes (10⁶ bytes) while Windows reports binary mebibytes but calls them MB. The NIST binary prefix guide formalizes MiB, but the habit persists. When you calculate file size in MB for a software limit, check the docs. I keep a column for both in my sizing sheets.
If you want to skip the manual division, our File Size Calculator outputs both conventions side by side so you never misreport to a client.
Image Files: Pixels, Bit Depth, and the Compression Gap
Let’s apply the formula to a 24-megapixel photo (6000 × 4000) at 24-bit color. Dimensions = 24,000,000 pixels. Bits = 24,000,000 × 24 = 576,000,000 bits. Bytes = 72,000,000. Binary MB = 72,000,000 ÷ 1,048,576 ≈ 68.7 MB. That is the uncompressed BMP/TIFF size.
Yet a JPEG from that camera is often 6–12 MB. Why? Lossy compression throws away perceptual redundancy, and the container adds only a few KB of metadata. The theoretical size is a ceiling, not a prediction.
Metadata and Containers Most People Ignore
When I first exported TIFFs for a gallery, I forgot that Photoshop embeds ICC profiles, thumbnails, and EXIF. A 68 MB raw pixel store became 71 MB on disk. Small in one file, but across 10,000 images that is 30 GB of hidden overhead. Always pad estimates by 3–5% for metadata.
Also, modern formats like HEIC use variable compression. A 12-bit RAW from a mirrorless camera stores Bayer data, not 24-bit RGB, so its uncompressed equivalent is closer to 34 MB, not 68 MB. Your calculated size is useless for final deliverables unless you know the true stored bit depth and quality factor. Use the formula for planning, then measure actuals.
Audio Files: Sample Rate, Channels, and Codec Reality
Take a 3-minute song (180 seconds) at 44.1 kHz, 16-bit stereo. Dimensions = 44100 × 180 × 2 = 15,876,000 samples. Bits = 15,876,000 × 16 = 254,016,000. Bytes = 31,752,000. Binary MB ≈ 30.3 MB. That is the WAV/AIFF size.
But an MP3 at 192 kbps is 192,000 × 180 ÷ 8 = 4,320,000 bytes ≈ 4.1 MB. The formula’s ‘bit depth’ for lossy audio is effectively the bitrate divided by sample rate, a variable number. Most people don’t realize that bitrate is already a compressed bits-per-second measure, shortcutting the dimension math.
Why Your WAV Is Larger Than Expected
Beyond PCM math, audio containers add seek tables and tags. A 30 MB WAV might show 30.4 MB in Explorer. On a FAT32 card with 32 KB clusters, it occupies 30.5 MB due to slack. Multiply by 1,000 tracks and you lose a CD’s worth of space to rounding.
High-resolution audio (96 kHz, 24-bit, stereo) blows up fast: 96000 × 180 × 2 × 24 ÷ 8 = 103,680,000 bytes ≈ 98.9 MB for the same 3 minutes. I learned to ask clients about delivery spec before quoting archival space, or risk a 3× underestimate.
Video Files: Bitrate Math and Container Overhead
Video is where the universal formula bends but does not break. A 1080p (1920×1080) 24-bit frame is 6,220,800 bytes uncompressed. At 30 fps for 60 seconds, that is 11.2 GB. No one ships that. Instead, we use bitrate: a 10 Mbps stream for 60 seconds yields 10,000,000 × 60 ÷ 8 = 75,000,000 bytes ≈ 71.5 MB.
The catch is that bitrate already bakes in codec efficiency (H.264, HEVC). Your dimension-based calculation is only useful for raw capture cards or uncompressed DPX sequences. For delivery, trust bitrate × time, then add 1–2% for the MP4/MOV atom structure.
Variable Bitrate and Why Averages Lie
I once encoded a webinar with VBR to hit a 500 MB cap. The calculated constant-bitrate size was 520 MB, but the actual file was 610 MB because high-motion slides spiked bitrate. If you need hard limits, calculate with peak bitrate, not average. A 4K 60fps HEVC clip at 25 Mbps for 5 minutes is 25,000,000 × 300 ÷ 8 = 937,500,000 bytes ≈ 894 MB binary, but the same clip in ProRes 422 is over 8 GB because ProRes is visually lossless and near-uncompressed.
How Big Is a 25MB File? Tangible Real-World Scale
Users constantly ask, ‘How big is a 25MB file?’ and ‘How big is a 10 MB file?’ These are not abstract numbers; they map to everyday items. A 10 MB file is roughly 3–4 high-quality JPEG photos from a smartphone, one minute of CD-quality audio, or a 10-page PDF with light graphics. A 25 MB file holds about 8–10 such photos, a 3-minute MP3 album track at high quality, or a 5-minute 720p video clip.
To make it concrete: if you email a 25MB attachment limit, you can send around twenty 1.2 MB phone pictures but only one uncompressed 24-megapixel TIFF (which would be blocked). Understanding this prevents failed uploads. When I set up a client portal, I used these anchors so non-technical staff could visualize limits.
For storage planning, 25 MB is trivial on a 512 GB drive (about 20,000 such files), but on a 128 MB legacy flash card it is 20% of capacity. Scale matters. A 10 MB plain-text file is about 5 million characters—roughly 2,500 pages of a novel—because ASCII uses one byte per character. If you are modeling backups, our Backup Size Calculator lets you sum thousands of these units without manual error.
OS File Properties vs Your Calculated Size
The most common support ticket I receive: ‘Your formula says 68 MB but Windows says 71 MB on disk.’ This is the filesystem cluster effect. NTFS default cluster is 4 KB. A 68.1 MB file actually consumes 68.104 MB of clusters because the last partial cluster is padded. Multiply by millions of files and the ‘size on disk’ can exceed ‘size’ by 5–10%.
Another hidden factor is filesystem metadata: inodes, journaling, and extended attributes. On macOS, a file with resource forks stores extra data invisible to simple byte counts. The thing nobody tells you about is that cloud storage like S3 also charges for metadata and request overhead, so your 25MB object costs more than 25MB of bandwidth. APFS on modern Macs can perform transparent compression, making a 10 MB text file occupy 3 MB on disk without you knowing.
Reading Properties Correctly
When you determine the size of a file, right-click and read both ‘Size’ and ‘Size on disk’ on Windows, or use stat on Linux for exact bytes. Never trust the rounded megabyte label in a thumbnail view. I train junior techs to always record the byte count, then convert, to avoid the binary/decimal drift.
The File Type Cheat Sheet and Step-by-Step Method
Below is the comparative guide I wish existed when I started. It unifies the media types into one table:
| Media | Dimensions term | Bit depth | Uncompressed MB formula | Typical compressed real-world |
|---|---|---|---|---|
| Image (still) | width × height px | 24-bit RGB | W×H×24÷8÷1,048,576 | JPEG 1/10th of calc |
| RAW photo | width × height px | 12-bit Bayer | W×H×12÷8÷1,048,576 | Lossless 1/2 of calc |
| Audio (PCM) | sample rate × sec × ch | 16-bit | SR×s×ch×16÷8÷1,048,576 | MP3 1/8th of calc |
| Hi-res audio | sample rate × sec × ch | 24-bit | SR×s×ch×24÷8÷1,048,576 | FLAC 1/2 of calc |
| Video (raw) | W×H×fps×sec | 24-bit | W×H×fps×s×24÷8÷1,048,576 | H.264 1/150th of calc |
| Text/PDF | characters | 8-bit ASCII | chars×8÷8÷1,048,576 | ZIP often 1/2 to 1/3 |
Step-by-step to calculate any file size in MB:
- Identify media type and list its dimensions (pixels, samples, frames, characters).
- Note bit depth or bitrate from source specs.
- Multiply dimensions × bit depth × time to get bits.
- Divide by 8 for bytes.
- Divide by 1,048,576 for binary MB or 1,000,000 for decimal MB.
- Apply compression ratio estimate only after noting the uncompressed ceiling.
- Add 3–5% metadata and cluster padding for real on-disk size.
This process has saved me from overselling storage to clients and from under-provisioning capture buffers on film sets.
Common Mistakes and Edge Cases I’ve Learned the Hard Way
Even with the formula, pitfalls remain. Here are the ones that bitten me:
Mixing MiB and MB in Contracts
I once signed a deliverable spec for ‘500 MB maximum’ using binary math, but the client’s Linux server used decimal. My 524 MB binary file was ‘over size’ and rejected. Now I write ‘MiB’ explicitly in contracts.
Ignoring Variable Bitrate and Lossy Codecs
As noted in video, averages lie. For audio podcasts, VBR can swing 30%. Always calculate with peak rates if the limit is strict.
Assuming All Pixels Are Equal
Some cameras use 12-bit RAW with Bayer mosaic; the file is smaller than 24-bit RGB because demosaicing is done later. Your dimension math must use the actual stored bit depth, not the displayed image depth. This edge case tripped me on a drone footage audit where I predicted 2 TB and the cards held 900 GB.
Small Files in Cloud Storage
When I migrated 50,000 10 KB JSON files to object storage, the nominal 500 MB became 1.1 GB after per-object metadata fees. The formula calculates payload only; real-world overhead can double tiny-file footprints.
Putting It Together: Plan Storage Like a Practitioner
Calculating file size is not an academic exercise. It drives backup cycles, email limits, and hardware purchases. Start with the universal formula to set the ceiling, then measure actuals on a sample set to derive your real compression ratio. Pad for metadata and cluster slack.
When estimating a year of CCTV footage or a music library, scale the per-file math by count, but validate with the Backup Size Calculator to include growth and versioning. The goal is not perfect prediction; it is avoiding costly surprises.
After a decade of writing imaging pipelines and audio archives, my rule is simple: calculate to understand, measure to confirm, and always distrust a rounded megabyte.