I've seen a couple of tests of actual real-world data transfer rate tests that answer your question. Test used the same actual drive in an external enclosure, and compared data transfer rates for copying a complete drive from internal to external units. The manipulated variable was the type of connection between enclosure and computer - USB2, eSATA, Firewire 400 (the common one, aka IEEE 1394a), or Firewire 800 (new IEEE 1394b). USB 2 was slowest among them, Firewire 400 a bit better because it wastes less overhead time on bus management. Data rate using eSATA varied - at its slowest on a full HDD it matched USB2, but at its fastest on a near-empty HDD it was twice as fast. Firewire 800 (not very common yet) was fastest of the bunch.
Note that the details of the HDD itself in the external enclosure were not a real issue. You can get enclosures for SATA, SATA II, or EIDE HHD's. Virtually all of them have faster data transfer capabilities than the enclosure-to-computer interface, so that detail may not matter. In my own case, though, I chose a SATA II HDD to put into en external enclsure just to preserve future possibilities for how that HDD is used. The enclosure I chose has eSATA (I am using) plus USB2 (for compatability with others) interfaces.
By the way, eSATA is slightly different from SATA - it includes a couple of features that may or may not be included in a SATA controller. And to ensure there is no confusion, eSATA does use a slightly different connector. For that reason it is better to get a real eSATA controller (my mobo has one built in). Some enclosure makers will include an adapter that lets you use a normal SATA II controller connector to hook up to an eSATA cable and I understand they work fine for most requirements; you just may not get all the features of the full eSATA spec that way.