To add to what others have already said, I should stress again that the nameplate voltage is assuming full load with a resistive load. This is a very common pitfall for people designing power supplies for the first time, and I've even seen similar mistakes in commercial gear. When you put a rectifier and capacitor on the output of the transformer, the peak current is very high - so the voltage can actually drop below the nameplate voltage. If you look at the transformer voltage on a scope, you'll see the peaks of the sine wave have been lopped off by the heavy current pulses charging the filter cap.
I don't know what the best way to calculate the voltage droop is - so far, my preferred method is to use circuit simulation. I calculate the virtual impedance of the transformer by subtracting the off load voltage (RMS rating * transformer regulation) from the nameplate voltage. Then, I divide that by the maximum RMS current rating of the transformer. You can then simulate the circuit by using an AC source in series with a suitable resistor. The peak voltage of the AC source should be RMS rating * transformer regulation * 1.4. But you also need to account for mains voltage tolerance (+/- 10%), so multiply that by 0.9 to get a worst-case minimum voltage.
Then you have additional sources of voltage drop - the rectifier diodes, and the ripple. I usually assume 1V peak voltage drop for a silicon rectifier (not schottky), so a bridge could lose up to 2V. The ripple might be 1 or 2 volts peak to peak (depending on the size of the capacitor), so there's another 1-2 volts gone. Then consider the minimum dropout voltage of your regulator (3V for a 7805), and you'll have to use a much, much higher input voltage than you might first expect. A 9V AC RMS transformer seems to be suitable for a 5V regulated power supply taking this into account, using a bridge rectifier and a 7805. A 6V AC transformer certainly isn't sufficient (even without rectifier voltage drop, ripple and transformer droop, the maximum peak voltage with -10% mains is only 7.6V).
The worst-case maximum input voltage will be the RMS voltage rating * transformer regulation * 1.4 * mains tolerance (1.1), then minus the minimum voltage drop of the rectifier diodes (0.6V for a full wave rectifier, 1.2V for a bridge). Your capacitor and regulator needs to be able to handle it.
I usually calculate the regulator dissipation by calculating the voltage drop across the transformer. To do that, I multiply my RMS load current (less than the transformer's rated current) by the transformer's virtual impedance (calculated earlier). Note that the transformer's AC RMS load current will be about twice the DC output current with a bridge rectifier. Having calculated the RMS voltage drop, I then subtract it from the transformer's highest RMS output voltage (RMS voltage rating * transformer regulation * mains tolerance). Subtract the minimum voltage drop of the rectifier diodes - and half the ripple voltage - and you can then determine the typical average input voltage to the regulator under maximum load. Subtract the output voltage of the regulator and multiply by the DC load current to determine power dissipation.
Finally, I calculate the required heatsink by dividing the allowable temperature rise by the wattage dissipated in the regulator. Allowable temperature rise needs to stay below max junction temperature (typically 125C) at the maximum expected ambient temperature. So for a 35C ambient, you can tolerate a 90C rise. Then, you subtract the junction to case thermal resistance from the calculated C/W rating of the heatsink. For a 7805, the junction to case is 5C/W. Then you have to subtract the case to heatsink thermal resistance, which could be as much as 1.4C/W if you don't use thermal paste. You might be surprised by how large a heatsink you need!
Take these calculations with a grain of salt, because I could well be wrong. I'd appreciate the input of anyone more knowledgeable in this area if there are any errors!
If you don't have access to the datasheet of the transformer, then I'm afraid it's going to be more difficult. You'll have to measure the incoming AC mains voltage, output voltage of the transformer under no load, and calculate the regulation from that. I've seen some small transformers with regulation nearly as bad as 60% - so it's not an insignificant factor!