There is typically about 5mg. A broken bulb in a relatively small room can easily result in airborne levels that are well in excess of established safe levels. You can minimize this with prompt clean-up and ventilation, but those of us with small children may not be aware immediately. This isn't a theoretical problem; I've personally found my children had broken a bulb in their bedroom, which makes me wary of using CFLs in any non-covered fixture.
An incandescent bulb does not contain any mercury. The numbers for mercury pollution with incandescents is strictly indirect, in that manufacture and operation require electricity and the majority of electricity generated in this country is from coal-fired plants which in turn release mercury.
There are, however, a few problems with that comparison. For starters, the estimates released by the EPA only account for the amount of mercury directly present in the finished products and that indirectly released during a bulbs estimated operational lifetime; they do not attempt to account for energy expended during manufacture, not do they measure pollutants released by the mining or materials or manufacture of components (such as the electronics require in a CFL).
They also do not reflect advances in technology or new regulations; for example, the GAO reports
an average 90% reduction in plants employing recent mercury control technologies. And of course any shift in the balance of energy production to cleaner sources reduces the indirect impact of consumption.
There is also a difference between pollution generated by a power plant, which will typically be dispersed over a broad area and the potential for localized contamination.
Are CFLs a net win? Possibly. Maybe even probably. But advocates do a disservice by focusing only on the superficial impact of technologies.