Mbps vs. MB/s: A Complete Guide to Speed Unit Conversion and Broadband Line Selection

The Difference Between Mbps and MB/s Is the

Difference Between Bits and Bytes

Mbps stands for megabits per second and expresses, in millions, the number of bits that can be transferred per second. MB/s stands for megabytes per second and expresses, in millions, the number of bytes that can be transferred per second. Both describe the same “amount transferred per second,” but they differ decisively in whether the thing being counted is bits or bytes.

One byte is eight bits. Therefore, when expressing the same line speed, a figure given in MB/s is one-eighth of the figure given in Mbps. A 100 Mbps line has a theoretical transfer capacity of 12.5 MB/s. Conversely, a transfer capacity of 50 MB/s is equivalent to 400 Mbps.

The biggest pitfall in notation is the distinction between lowercase b and uppercase B. Lowercase b means bit, and uppercase B means byte. Japanese telecommunications carriers tend to use Mbps notation, while operating systems and storage products tend to use MB/s notation. There is no end to cases in which people compare figures from different measurement bases as they are and misjudge a line’s capability.

Why Two Types of Units Coexist

In the design of communications technology, there is a history of managing line speeds in bit units. Physical-layer specifications such as modulation schemes and frame structures assume bit strings, and bit units were natural when describing transmission-path capacity. ITU recommendations and IEEE 802.3, among others, also use bits per second as the basic unit.

On the other hand, file sizes and memory capacities are handled in byte units. OS file-management screens and storage product specifications standardly use byte-series notation, and calculating download times in byte units is practical. These two conventions collide at the point of connection, producing conversion misunderstandings.

In addition, the definition of prefixes is also unstable. The SI prefix mega means 10 to the 6th power, but in contexts premised on binary notation it can mean 1,048,576. To make this binary prefix explicit, the International Electrotechnical Commission specified the notation MiB (mebibyte) in IEC 80000-13. Decimal prefixes are dominant in communication speed notation, but a mixture remains in storage capacities.

Conversion Formulas and Quick Reference Table

Conversion can be done with simple division and multiplication. To go from Mbps to MB/s, divide by 8; to go from MB/s to Mbps, multiply by 8. The correspondence for major values is as follows.

• 10 Mbps = 1.25 MB/s
• 100 Mbps = 12.5 MB/s
• 200 Mbps = 25 MB/s
• 500 Mbps = 62.5 MB/s
• 1 Gbps = 125 MB/s
• 2 Gbps = 250 MB/s
• 10 Gbps = 1,250 MB/s

When downloading a 10 GB file over a 1 Gbps line, if you calculate 10 GB as 10,000 MB, you get 10,000 ÷ 125 = 80 seconds. However, due to the drop in effective speed described later, it actually takes several times as long. In reverse calculations, the procedure of deriving bandwidth from the required transfer time is useful. If you want to send 600 MB of data in 3 minutes, 600 ÷ 180 seconds = 3.33 MB/s, so an effective speed of 26.7 Mbps or higher is required.

Factors Behind the Gap Between Catalog Values

and Effective Speeds

Figures such as 1 Gbps presented by line operators refer to the theoretical maximum at the physical layer or the upper limit of a best-effort service. Measured values will be lower than this. The main factors can be organized into four.

First, protocol overhead. TCP/IP headers, Ethernet frame headers, and PPPoE encapsulation are added. With PPPoE connections, up to about 8% of bandwidth is consumed. Switching to IPv4 over IPv6 can reduce this encapsulation load and may improve effective speed during congested hours. However, communication to destinations that do not support IPv6 continues to go through IPv4 routes, so not all communication becomes faster.

Second, control by upper-layer protocols. TCP window control, retransmission processing, and TLS handshakes have an effect. When communicating with a server that has high latency, the window size becomes a constraint, and speed does not increase even if there is spare bandwidth. Third, the performance of in-home devices. A router’s NAT processing capability, Wi-Fi effective throughput, and the LAN port standard determine the upper limit. The reality is that the effective value for Wi-Fi 5 is on the order of a few Mbps, and even Wi-Fi 6 does not reach several Gbps. Fourth, measurement conditions. Results vary greatly depending on distance to the measurement server, congested time periods, and the number of simultaneously connected devices.

Estimating Required Bandwidth for Choosing a

Connection

When choosing a line, a realistic method is to add up the required bandwidth for each use rather than focusing on the size of the contracted bandwidth. Video streaming services publish their recommended bandwidth. Netflix advises 5 Mbps for HD quality and 25 Mbps for 4K. Zoom advises about 1.5 Mbps each for upload and download in group calls and 2.5 Mbps or more for high-definition settings.

Assuming simultaneous use within a household, add them up as follows.

• Simultaneously watching two 4K videos: 25 Mbps × 2 = 50 Mbps
• Two telework video conferences: 2.5 Mbps × 2 = 5 Mbps
• One online game: 10 Mbps
• Set a headroom factor of 2: total 65 Mbps × 2 = 130 Mbps

For this usage pattern, a line that delivers an effective 100–200 Mbps is sufficient. The significance of a 1 Gbps contract lies not in single-session speed but in simultaneous use by multiple devices and short-time transfer of large files. Note that for online games and remote operations, latency and jitter have a greater impact than bandwidth. There are areas where quality cannot be judged by speed figures alone.

Estimating Transfer Time for Business Use

In cloud migration and backup planning, the time required for the initial transfer affects the entire project. When transferring 500 GB of data over eight hours at night, 500,000 MB ÷ 28,800 seconds = 17.4 MB/s, so an effective speed of 139 Mbps or higher is required at all times. Because effective speed varies by time of day, this calculation must be multiplied by a safety factor.

Differential backup and synchronization tools behave differently for the initial full transfer and for the second and subsequent differential transfers. The decision to augment a line should be evaluated based on measured values for the initial transfer and the daily amount of change. When using a VPN for inter-site connections, it should not be overlooked that encryption processing and tunnel overhead make the router performance on the site side a new constraint.

How to Interpret Speed Measurement Results

When interpreting measurement results, check the following points.

• Check whether it is a single-session measurement or an aggregate of multiple sessions. Ookla Speedtest has a setting that uses multiple connections in parallel, and an aggregate value may be displayed.
• Check the displayed unit. Mbps and MB/s are mixed depending on the app or service.
• The closer the measurement server, the higher the result. A measurement on a domestic server does not guarantee effective speed with an overseas server.
• Evaluate wireless connection measurements by comparing them with wired connection results. If the difference is large, the in-home wireless environment is the constraint.
• Measure multiple times at different times of day and understand the range of variation.

Notes on Notations That Are Easily Confused

• “1 Gbps is 1 GB/s” is incorrect. 1 Gbps is 125 MB/s.
• “The faster the line, the more comfortable” is not necessarily true. On lines with high latency or packet loss, the quality of calls and operation response degrades even if bandwidth is wide.
• “Wi-Fi standard values are synonymous with effective speed” is not true. Standard values are theoretical maximums, and effective values are generally half or less.
• “MB/s and MiB/s are the same” is not true. The former is based on decimal prefixes and the latter on binary prefixes. The difference between OS displays and product specifications originates here.
• “Uplink and downlink speeds are the same” is not necessarily true. Many fiber-optic lines are symmetric, but cable TV-based and mobile lines prioritize downlink.

Editorial Opinion

As an evaluation axis for comparison, the editorial team believes that effective throughput and the stability of latency and jitter should be checked separately, rather than the maximum catalog speed. In particular, for lines aimed at apartment buildings, congestion during hours when the same equipment is shared affects perceived quality, so a judgment cannot be made based solely on the size of the contracted bandwidth. It is effective to add up the required bandwidth for each use and compare measured values across multiple time periods.

A pitfall in the field is that in-home devices and the wireless environment become bottlenecks for the contracted bandwidth. Even with a 1 Gbps contract, if a router with only 100 Mbps LAN ports is placed in between, the speed hits a ceiling there. There are also many cases in which people compare figures without checking the display unit of the measurement app or whether parallel connections are used.

As for future direction, the editorial team assesses that with the spread of 10-gigabit-class services, the importance of device-side reception capability and LAN standards will increase. At the same time, the question of whether speed notation should lean toward bits or bytes remains, and the focus will be on whether an operational practice in which both operators and users explicitly state units takes hold.

References

• IEC 80000-13:2008 Quantities and Units — Part 13: Information Science and Technology
• IEEE 1541-2002 Standard for Prefixes for Binary Multiples
• Netflix Internet Connection Speed Recommendations
• Zoom System Requirements
• Ookla Speedtest Official Site