You've probably already read quite a bit about the selegiline (L-deprenyl) patch (right?), but these four (five) points may frame the information more usefully.
1a. All the oral MAOIs you are used to (phenelzine, moclobemide, trancylpromine, etc) are either nonselective (both MAO-A and MAO-B inhibitors) or are selective MAO-A inhibitors.
1b. MAO-A inhibition is needed for antidepressant effect.
2. MAO-A metabolizes serotonin, norepinephrine, dopamine, and tyramine.
3. MAO-A in the gut is what metabolizes tyramine. Inhibition of the gut MAO-A allows tyramine to enter the circulation unmetabolized-- thus releasing norepinephrine and causing hypertensive crises.
4. Oral selegiline (pill) is an MAO-B inhibitor at doses less than 10mg/d.
In other words, a) selegiline requires no dietary restriction below 10mg/d (because it doesn't affect MAO-A in the gut) and b) it doesn't work below 10mg/d (for depression; MAO-B metabolizes dopmaine, so selegiline will still be good for Parkinson's at small doses.)
5. Above 10mg/d, selegiline is nonselective (thus MAO-A and B inhibition). Thus, a) it should work; b) it will require dietary restrictions.
One interesting point: selegiline is rapidly metabolized (first pass) to desmethylselegiline, l-amphetamine, and l-methamphetamine. (1)
The point of the patch is that it bypasses the first pass metabolism (you don't eat it) so you get much higher concentrations of drug into the CNS and few metabolites. Also, much less goes to the intestinal MAOs, so you get both MAO-A and B inhibition in the brain, but less of the MAO-A in the intestine. So even if you use doses greater than 10mg/d, you (probably) don't need dietary restrictions. (NB: even though I can't find any studies clearly linking the risk (most find it safe up to 20mg) the PI still says to avoid tyamine foods above 9mg/d.)
Part 2: Efficacy
Above, I made the outrageous statement, "it doesn't work below 10mg/d." What's really outrageous is that I couldn't find any evidence that it worked above 10mg/d, either.
Here's a typical example: a 2003 study of 289 patients, double blinded, placebo controlled, of selegiline patch 20mg/d (keep in mind, the starting dose is 6mg/d) vs. placebo patch. Though the paper finds "statistical superiority" of the patch over placebo, it took 8 weeks to get a 2-3 point difference on the MADRS or HAMD-28. (For context: the HAMD-28 has 28 questions with ratings from 0-4. So three points difference could be three points on one question, or one point on three questions...) It never beat placebo on the HAMD-17. (To the author's credit, he does not hide this and is upfront that these were "modest" differences.)
Contrast that with the first clinical trial of the selegiline patch (done, astonishingly, by the same author): superior efficacy on all three scales. But, of course, even at 20mg/d, it's really not that superior:
Maybe 2-4 points, max? I grant more people responded to the patch (as defined by reduction of 50% on the HAM score)-- but it was 15% more people, and, well, come on on...
Just to make the point, a 67 person, multicenter, double blind, placebo controlled study tested oral selegiline's efficacy in schizophrenics, and found improvement vs. placebo in "negative symptoms," as defined by the Scale for the Assessment of Negative Symptoms (SANS). Troublingly, "improvement" means one point difference:
And not much happened for depression (HAM-D) either.
Someone, somewhere is going to accuse me of only showing weak studies and omitting all the studies that showed it worked well. Okay. Here is the last known study:
The only other patch study was a 321 person, 1 year long placebo controlled study, found that while twice as many people dropped out for side effects (13.2% vs. 6.7%), twice as many on placebo relapsed at 6 months (16.8% selegiline relapse vs. 29.4% placebo). Interestingly, at one year the relapse rates for both drug and placebo were identical-- in other words, all relapses occurred in the first 6 months, none in the second 6 months.