I have already made most of my travel arrangements for the APS March meeting, which I am attending in 2008. I used to go every year until maybe ten years ago; now circumstances are different, and on average I go every 3 to 5 years. I remember the time when most condensed matter physics faculty would attend this conference annually or bi-annually, typically taking along with them graduate students and postdoctoral associates. The same was true for APS meetings in other areas of physics.
The main reason why going every year has become a difficult proposition is its sheer cost.
On making housing arrangements ahead of time, one can get a room at one of the conference hotels, typically high end ones. Very seldom will one be able to pay less than $100/night, even by sharing a room, and more likely much more than that, especially if one postpones the decision of attending until the last moment, like many of us do. Expenses might be reduced somewhat on staying at a motel, but in my experience this option is not worth pursuing. For one thing, typically it will mean forgoing the proximity to where the conference takes place. This is inconvenient, as a conference (especially one like an APS meeting) is normally attended also/mostly for networking purposes, which requires talking to many different colleagues, especially over dinner. Moreover, almost inevitably a car must be rented, driven every day to the conference and back, parked and refueled, which can substantially reduce the save of money over staying at one of the conference hotels, from which one can just walk or take a shuttle bus to the conference.
To that one must add the conference registration fee, typically few hundred dollars, as well as airfare, also worth several hundred dollars, and at least two meals a day, almost inevitably consumed at pricey restaurants within walking distance to the conference.
Bottom line is, we are talking at least $1,500, and much more likely over $2,000 per attendee. Annual travel allowance for a faculty based in North America is typically around that much, which means that one will likely have to foot part of the bill out of pocket, and then be left with no travel money to attend any other conference.
Now, for most faculty paying out of pocket is not a problem, and what we get out of attending is well worth what we pay. The real burden is on graduate students and to a lesser extent postdoctoral associates. These are arguably the ones who benefit the most from attending a conference, for a number of educational and professional reasons, but also the ones who can afford it the least. A graduate student living off 20K/year can hardly sustain comfortably an expense of this magnitude .
Where can the money come from, then ? Often funds are available from various local sources, such as large group grants, university-based research centers and institutes, possibly some targeted scholarships, but my sense (I have no data) is that the majority of graduate students attending the conference are doing so on the grant of their major professor. Using grant money, most faculty will be able to help one student, but certainly not four (my current case) . Even if it is easy to explain that senior graduate students should be given priority, there may still be more than one of them, and ideally it would be nice to take the whole group along. But unless APS decides to give a break to students and make it financially easier for them (waiving registration fee and providing cheaper housing arrangements are the first things that come to mind), and/or universities are willing to allocate special funds for this purpose, ideally paying upfront for things like air fare, so that no actual money has to come out of the student’s pocket, I am pessimistic about future student attendance of APS flagship meetings.
Perhaps emphasis will shift toward smaller regional meetings, held at university campuses. In a way it’s unfortunate, because for a student attending a large international conference like an APS meeting is in many respects a remarkable experience.
 As Female Science Professor has pointed out, even when the money is provided, it can be a while before expenses are reimbursed, which makes it difficult for someone at that income level.
 Things do not appear likely to improve any time soon. See, for instance, this posting on the Computing Research Policy Blog.