Hi,
I experienced same on some of my Cyclone and CycloneII boards. Everytime this happened, some programmings and "manual starts" later the board (in fact the FPGA) died completely. I did exchange the FPGA on some boards and the EEPROM was not affected - board operated. I think there is something with the nConfig signal, perhaps the additional resistors in the Programmer "helps to activate" the configuration process.
Second thing is, that this problem occurred only on boards where I programmed my EPCS with just connecting the USB-Blaster w/o disconnecting power first. On boards the programmer is connected with the power (of the to be programmed unit) been shut-off first, I did not observe this (and I did also a lot of reprogramming on "those boards").
Maybe there is a ESD problem as there are different voltages on the USB Blaster (or programmer) interface referenced to PC "GND" and the boards power supply.
For the Cyclone III there is an Application Note referring the protection diodes. I also came across an article suggesting the Programmer interface's GND line first...
Carlhermann