Didn't even know they existed. If they're cheaper than Analog parts, it might be interesting, but I like how Analog puts out a reference signal and you can change the response range by varying a few external passives. Anyway...
You've got 3 analog lines coming off that sensor, for your X, Y, and Z axis. There are no "center" signals for comparison, so you have to depend on your calibration. They provide two digital inputs, the PS pin for shutting down the unit and the TEST pin that shoves all accelerometer readings up 1G for calibration. Obviously, you only REALLY need the three analog lines.
Now, your measurement range is 0.33V to 2.97V. Not too shabby, you probably won't need to level shift (it is a 3.3V part after all). I'd have to check the PIC datasheet to see what the recommended analog Z is, but I think it's 2.5Kohm max, so you might need to throw in an op-amp voltage follower circuit for each analog line. If you don't, the ADC might read low to not being able to gather enough charge for an accurate reading.
Hmmm... I see the sensor will work at up to 5V. Appears that the unit is ratiometric, so... *scribble scribble scribble* at 5V you're looking at a center of 2.75V, 1V/G, so your range then would be 0.75V to 4.75V. This is actually a bit of an advantage, as you don't have to worry about varying power supply levels (battery) affecting the reading
So, summary, I'd personally run each of the three analog lines into a voltage follower circuit then to the PIC's ADC. If it's a low power design, hook up the PS and the TEST digital lines, otherwise, TEST alone should be sufficient to calibrate the unit. A callibration routine of placing it on a level surface may be recommended, too
