> Is it though? I think it is. That's certainly my experience with CS. You need some combination of GPA, recommendations from known researchers in the field, r

> Is it though? I think it is. That's certainly my experience with CS. You need some combination of GPA, recommendations from known researchers in the field, research experience/accomplishments and a strong undergraduate school. You don't need all of these always, but they all have an effect. Most of the grad students I knew at Berkeley CS came in with at least one good publication and high GPAs. Similarly, some very capable people I know from less-well-known universities (in the US and internationally) were rejected from most programs they applied to. My observations combined with the low % acceptance rates lead me to believe it's definitely difficult—especially if you come from an even slightly non-standard background. And this is in a field where most people don't pursue PhDs—I don't even want to think about what Math and Physics programs must be like. On the other hand, you can get some pretty deep (and well-paying!) positions in industry without having a degree at all, and people won't even look at your GPA. It's a lot more flexible.