First, its likely the unit proposed is 12,000 BTU heat and cool, not 1200 so its significantly larger capacity than your previous window unit. Most mini-splits are much more efficient than your window unit so energy savings will be in order, not to mention its a heat pump replacing strip heaters. 12k may be too large, 9000 units are available and cheaper. Also efficiency up to a SEER of 19 and they are VERY quiet. I would stick with a major brand unit, not a chinese import with no parts support. LG, Fujitsu, Mitsubishi, Panasonic and Sanyo. Unfortunatly the american brands have not really entered this market yet, they are much more popular overseas.

The questions from tigerdunes are important to answer. The cost of adding heat might be small, but it does add some mechanical complexity. Reversal valves in heat pumps have been known to fail. I have another question about your heat. Is it zoned? If not, you can heat just the downstairs. That said, heating just the downstairs tends to be difficult since hot air rises.

All true, Austin. My first objection was to the post that appeared to say that you lose any setback gains when you are in the recovery mode. You post appeared to accept that and I found it hard to believe that you do given what you have written in the past. That is a really annoying urban legend that I believe can be traced to early heat pump installations with very simple controls that could not recover from any really heating setback without hitting the backup, electrical resistance heat strips. That was fixed, obviously with better thermostats. Things they are a'changin, however. One important point is that with variable speed systems, maximum output puts you in relatively inefficient operation. That has led some writers to advise homeowners similarly, don't do setbacks. In fact, that is too simple. If the setback is deep enough and of long enough duration, the building characteristics lead to high loss, and you are not too close to the outdoor design temperature, setbacks can still be useful. The trouble is that most homeowners are not able to make the judgement. Many HVAC pros probably can't do it either. The fact is, to determine what will work best requires the kind of calculations that HVAC engineers are used to making and some good starting data. That is why the cooling equipment makers of the HVAC industry needs to get on the stick and catch up with the boiler makers (and probably furnace makers) who have relatively elegant controls allowing efficient setback recovery at partial output. They make it work, so can the cooling equipment gurus. Eventually, we should be able to get to the point where someone can look at their thermostat, or home computer, and see a message from the system telling us that it has decided that due to the weather forecast, it is not doing a setback today because it won't be able to return to your comfort level (conditions too close to outdoor design temperature). Those heating systems (mod cons) can take advantage of require considerable set-up and tweaking although I think that they are getting better at self-learning. Some are so adaptive that not much or any learning is needed to operate pretty efficiently. If anyone is not familiar, here is an example. I am making up the concrete numbers, but they are probably not too bad: The boiler starts up to recover from a setback at 60% output. The controls know the outdoor temp and the target temp and have already determined what time to start the recovery to have the floors warm by the time pop usually gets up (before the kids and wife) to get to work. Uh oh, the temperature rise is not as expected after 30 minutes so it bumps output up to 80%. After another half hour, still not good. It is a little too late for dad since he is leaving a little earlier than usual today, but it bumps output up to 95% and makes it time for mom's feet to hit a warm floor. (OK, efficiency has dropped from 97% to 92% for that last 30 minutes, but no biggie.) Cooling systems have the ability to look at outdoor temp and determine start-up time, but the modulation, on the fly or not, is not there for residential systems as far as I have seen. I'll be optimistic and believe that we will get there soon.

sjordan, not answering for klem, just my point of view: when cooling with interior units only in some rooms, other rooms will be cooled indireclty from normal heat/air exchange within the house. Also a lot of how we perceive heat is the amount of humidity in the air, and a lot of A/C is removing the humidity, which can benefit other rooms (and then how tight is the house in terms of penetration of humid air). Heat rises within the house, so you get houses where upper floor bedrooms can get stifling - but possibly only need A/C for a few hours a day, to remove heat before bedtime, are nights cool where you are, etc. Plus solar gain - south facing rooms may get uncomfortably hot, rooms that don't get sun be okay much of the day, shade effects. Air flow within the house, can fans and ceiling fans help. Etc. So, too many factors to generalize, your climate matters, and also depends on how demanding you are about temperature differences. If you focus first on rooms where you spend the most time (usually kitchen, main recreation room, occupied bedrooms), the rest may work itself out. Some households want 72f year-round, others are tolerant of quite a bit warmer during hot summers. The thing with at least some older houses - it was considered normal e.g. to have bedrooms quite a bit colder (more blankets!) or warmer (sleep with a fan, no sheets or blankets) in tune with the time of year. During summer, wasn't uncommon to have outdoor kitchens. Getting to the modern ideal of identical temperature everywhere may be a tall order - possible but a lot of work and expense (or lots of heating / cooling units).