The raw number isn't very meaningful. There are over 250 million passenger vehicles registered in the United States. That comes to one fire per 1641 vehicles. There are something like 12 thousand Teslas, but also only three fires, which comes to one fire per 4000 vehicles.
So with those numbers, they still seem better off than combustion driven vehicles. However, fleet size isn't the only variable. The majority of Teslas are less than a year old, while overall only 13.5% of passenger vehicles are under two years old, and 60.6% are at least seven years old. Teslas are also a luxury vehicle with a base price of $69,900 (before tax credits) and allegedly stringent manufacturing standards, compared to the overall market with an average new sale price of $24,764 and varying quality. And finally, maintenance needs to be taken into consideration. Not only are Teslas newer and almost certainly under warranty, the average owner is relatively wealthy and presumably can afford proper maintenance.
But even that is not necessarily the right set of stats to look at. The three Tesla fires occurred after an accident, so it may be more informative to look at the liklihood of accidents to result in fire. According to the NFPA, only 4% of automobile fires were the result of a collision or roll-over; most were caused by a mechanical or electrical failure. There were about 5.4 million accidents in 2010, so at 4% of 152300 accident it comes out to approximately one fire per 900 accidents. There are no hard statistics on precisely how many accidents there are for Tesla specifically, but it would have to be a significantly high number to have a lower fire to accident ratio. And that isn't even taking into account the relative severity of accidents into account.
Of course the real problem is that there simply isn't enough data available. Three times in as many months could be a fluke. Or it could be a pattern.