I understand why two computers need a crossover cable between themselves to communicate. Without it, the transmit and receive wires are the same on each end, making any communication impossible. What I don't understand is if I have a regular patch cable connecting a switch and host, like so:

enter image description here

The way I'm understanding this, the transmit/receive wires are still the same between two different devices. If the switch wants to send something to the computer, the computer will receive it on its transmit wires. How does a straight-through cable not run into this issue? Are switch/router/hub ports configured differently in that they have their transmit/receive pins swapped from the computer's?

  • 1
    Welcome to Network Engineering! You can find the answer to your question, and much more detail than we can provide here, by searching for Auto MDIX. That's the technology used to find the correct cable polarization
    – Ron Trunk
    Mar 2, 2023 at 17:14
  • Latest NICs can auto-negotiate and rearrange the wires if the auto-mdix feature is enabled. Mar 6, 2023 at 18:27

2 Answers 2


First, you must understand the DTE/DCE concept. A DCE (Data Communications Equipment), such as a switch, has the Tx/Rx lines reversed in its connector from a DTE (Data Terminal Equipment), such as a PC, so straight-through cabling has always worked for connecting a DTE to a DCE.

The problem arises when connecting a DTE to DTE or DCE to DCE. In older ethernet, crossover cables were required when connecting like devices. As Ron Trunk explained in his comment, you need to look at the subject of MDI/MDI-X and Auto MDI-X. Modern ethernet has built in Auto MDI-X capabilities where two devices connecting first discover and correct for the need of crossover.

The details of this subject are far too large for a site like this, so you will need to do some research.


Yes, each transmitter needs to connect to the far side's receiver.

Initially, twisted-pair Ethernet (as StarLAN) used already existing voice-grade cabling that used 1:1 connections. Accordingly, AT&T thought up a scheme were concentrators like repeater hubs or switches included the required crossover within their downlink port (router ports are MDI). Later, IEEE 802.3 called the end-node configuration MDI and the crossed-over concentrator configuration MDI-X. That way, connecting unlike ports (MDI⇔MDI-X) used the (then already common) "straight" 1:1 cabling. However, connecting like ports (MDI⇔MDI or MDI-X⇔MDI-X) required a crossover cable.

Some repeater or switches included a switchable port to be used as uplink to another concentrator with a straight cable. Some others used two complementary incarnations of the same port (which you could only use one of at any time).

Today, nearly all twisted-pair ports support Auto MDI-X that finds a match for the link partner automatically, so the need for a crossover cable or adapter has become very rare. 1000BASE-T and faster use all four pairs for simultaneous transmit and receive, yet (now mandatory) Auto Negotiation still requires an initial crossover to work.

In contrast, fiber cables are generally crossed since there were no legacy installations to deal with and all fiber ports are in MDI configuration.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.