The following is an attempt at describing both communication methods in brief.
With Token Ring, as the name suggest, a free token passes around the network. If a station wants to transmit, it receives the free token, discards it and replaces it with its own data frame. When a station then receives that data addressed to it, it marks the frame as received and passes it back out onto the network. The frame is received back at the originating station where it discards it and then releases a new free token onto the network. Token Ring devices are often quite a bit more expensive than the ethernet equivalent due to the cost of the extra logic the boards need to have as well as having to have MAU's or CAU's etc.
Ethernet uses the CSMA/CD specification which stands for "Carrier Sense, Multiple Access / Collision Detect". When a station wants to transmit on the network it first 'listens' to the cable until it detects a quite period where there is no traffic on the network. At this point it will try and transmit its data. If another workstation has also detected the quite period and transmits at the same time you get a collision, and both stations have to wait a random ammount of time before retrying. Waiting for a random time allows one of the workstatins to retransmit first.
Ethernet is available at speeds of 10, 100 and 1000mb/s. There are many tools, utilities and various suppliers of devices for ethernet, whereas the choices for Token Ring are much more limited.
Token ring was more resilient than a coaxial/BNC cabled ethernet network, but UTP with hubs and switches etc these days mean that the cabling can be much more fault tolerent. So assuming that when you talk of ethernet, you ARE talking about UTP structured cabling and not the older coaxial cabling which is prone to noise and failure (a break on one cable segment brings all stations down on that segment) ethernet should give more resilience and easier upgrade/expansion as well as higher speed options.
That, that is, is.
That, that is not, is not.
That, that is, is not that, that is not.
That, that is not, is not that, that is.
Very nice summary. I might debate the statement that ethernet is "more" fault tolerant than T-R; I'd say it's now as fault tolerant (T-R's fault tolerance was its big advantage over E'net back in the early days). But that would be nit-picking.
The number of options (things like Snap servers, cd-rom towers, tape libraries, comm servers, etc) for ethernet-based networks is orders-of-magnitude above what you can get to run on T-R. Not to mention the fact that most desktop PCs and many peripherals already come with an Ethenet NIC. You'd have to buy T-R adapters for each new machine, and compared to E'net, they ain't cheap!
Next, even if your T-R network is based on UTP, you can't get more ports by just adding a mini-hub to an existing drop (not cheaply, anyway!)
There is a new breed of T-R that offers 100 mbps speed, but this is just playing catch-up with Ethernet. Having a 100-node T-R network here, I can tell you without hesitation that if you have a choice, go with Ethernet (and if you're designing from the ground up, consider switched 100-mbps to the desktop; its not that much more expensive than using hubs).