I compiled the Metrorail service interruption data for May 2008. My method and categories are described below. The coded data is available here, and the spreadsheet is available here. For a full discussion, see this document. WMATA has stated that the information in this post is being reviewed for comment.
Without further adieu, the data:
Miles traveled between failures (more is better):
How does Metrorail compare to other transit systems? The data in the figure above is from the NYC Straphangers campaign (PDF). Based on some rough figures (6 car trains, 12 mile commute each way, 240 days per year, each delay affecting 2-3 trains in addition to the actual failed one), this means an average WMATA rider could expect to be on a train that is delayed 10 minutes or more three times per year. Compare that to NYC, where it should only happen once or twice per year on the worst line, and less than once per year on the average line.
Looking at it another way, the delayed or out of service trains are only 0.6% of all trains. But this means that you're likely to be delayed in one out of every 180 trips or so, or for the average commuter, about every 4-5 months. When a train is delayed, other trains in the system are also delayed, depending on the severity of the problem. If you assume that three or four other trains are adversely affected, then passengers might expect to experience one delay per month.
I think this reflects the kind of service I've experienced. I haven't experienced these kinds of delays in other heavy rail systems, but I will grant you that I ride about 200-300 times more trips on Metrorail than all other rail systems combined.
My brother rides BART for his daily commute and does not remember being offloaded ever. My conclusion is that Metrorail probably suffers from higher than typical service interruptions, based on data publicly available from WMATA's website. Hopefully, I will obtain data for other systems for comparison purposes. I'm not trying to be overly critical of WMATA operations here, they operate a lot of trains and most (92% [PDF]) of them are on time. My commute is affected only on an occasional basis, but it still seems like it's more often than in other systems with which I have experience.
This data is important to the debate about WMATA capital funding (PDF). On the one hand, the lack of capital funding could be contributing to decreased system reliability. On the other hand, system problems run the risk of being perceived as the fault of mismanagement or waste and could make it harder to obtain additional funds.