Jason, I have not much to say in addition to solutions offered. But, here is mine

- If you are able to to feed the selenoid from a seperate supply, do it. I am for Mahmood solution if the case is so.

- If you have to feed both the microcontroller and selenoid from a single supply,
first, measure the selenoid`s active state (holding state) current. This current is supposed to be 1 out of 5 of the initial state current (the drawn current at moment the selenoid has been activated) Lets say holding state current is 100mA under 12V. Then,put a 120 ohm resistor in serial with selenoid and put a large cap. (1000mF) in between GND and to the point the resistor and selenoid meets.
Large cap. will meet the initial current (momentaraly) and 120 ohm resistor will suffice to hold the selenoid at on state. This also allows the cap to be charged again for the next activation.
But if you are to control the selenoid`s very frequently, this method probably will fail just because the cap has no time to charge for the second trial.


But never forget the other measures to be taken as the other friends suggested.

Rustu