Not to beat this thing into the ground, but just to clarify where I was coming from. The book in question ("Psychopharmocology: An Introduction" by Rene Spiegel, if anyone is interested) classifies depression into three categories:

Somatic Depression
Depression that comes about because of some physical illness. The authors recommend neither therapy nor meds for this -- they just figure the doc needs to treat the underlying problem to fix it.

Psychogenic Depression
Depression due to some traceable psychological or life circumstance. The authors recommend therapy for this, only to be followed with meds if the therapy isn't working.

Endogenous Depression
Depression that doesn't result from either physical illness or obvious psychological issues. (The concept of a chemical imbalance falls into this category, too.) The authors recommend meds for this, to be followed with therapy if the meds aren't working.

I only mentioned it because I expected drug company researchers to recommend meds for everything, and I was surprised at how conservative they were. I think docs out in the real world have their own guidelines for prescription.