Does polling a pain = battery drain?

Forgive my stupidity in advance on this one...

Let's say I have an ATMega (or Arduino) running off a battery, and I have a program running on it that it is polling a pin constantly and kicks off some steps if the pin state changes.

How would the battery life on that setup compare to, say, a program that assigns an interrupt to that pin and then sits their idle until the interrupt triggers? Does the explicit pin polling suck on the battery faster than the waiting for an interrupt?

Thanks!
Matt

Does the explicit pin polling suck on the battery faster than the waiting for an interrupt?

Yes. I am working on a project to measure wind speed at the moment and with polling, I could not get much lower than about 500uA, but sleeping and waking on external interrupts, down to 15uA.

Thank you! Just what I needed to know.

It makes a major difference whether you let the processor "sit there idle until the interrupt triggers" or put the processor into sleep mode.

Yes it does, because in SLEEP_MODE_IDLE the main clock is running, and all the other parts of the MCU like timers, serial/i2c/SPI, ADC are still working. I am using SLEEP_MODE_PWR_SAVE, where everything is shut down except Timer2 and the ability to wake on external interrupt. SLEEP_MODE_PWR_DOWN is even lower, but I can't use that in my particular project because I need to time long intervals like 6secs or 15mins reasonably accurately, so I'm using a 32KHz crystal to run Timer2.