A commenter wrote the following: "About PRL, I do have a concrete question. You've been around for some time, I am new in the business. Can you explain what happened to it? Twenty years ago it used to be the journal to publish in, now it is an afterthought."
Physical Review Letters remains a premier place to publish high impact physics results in letter-format (that is, typically 4-ish page papers with around 4 figures). I think that the recently arrived editor in chief Pierre Meystre is working hard to revitalize PRL as a "destination journal" for physics results, where you know that the primary audience comprises physicists.
That being said, the origins of some of PRL's (possible) loss of cachet are immediately obvious. Twenty years ago, Nature and Science were not as highly regarded within the physics community as places to publish. Nature Publishing Group did not have all of its various progeny (Nature Physics, Nature Materials, Nature Nanotechnology, Nature Photonics being the four most relevant here). Likewise, the American Chemical Society's journal offerings used to be a lot less friendly to physicists. Now there are Nano Lett., ACS Nano, ACS Photonics, ACS Applied Materials and Interfaces. It's a far more competitive journal marketplace, and the Phys Rev journals have been slow to innovate. Furthermore, I think there is a broad perception that PRL's refereeing can be long, arduous, contentious, and distressingly random. Some of the competing journals somehow are able to be rapid and on-target in terms of the expertise of the referees. If you have a hot result, and you think that refereeing at PRL is highly likely to take a long time and require a protracted fight with referees, other alternatives have room to make inroads.
Somehow PRL needs to improve its reviewing reputation in terms of accuracy and timeliness. That, I think, would be the best way to be more competitive. That, and a re-design of their webpage redesign, which is neither particularly attractive or functional.