Good point! But that makes me think that maybe taking the average of a bunch of P/E ratios is mathematically stupid to begin with. I mean, you could have one with earnings of .00000001 which would make the P/E ratio enormous, or with earnings of -.0000001 which would make the P/E ration enormously negative. Or you could have earnings of exactly zero which would give you a NaN. So maybe this is just is an indication that the median P/E is a much more meaningful quantity than the average. Maybe taking the average of a bunch of ratios is meaningless, in general. Or maybe they are adding up total price of all stocks on the exchange / total earnings of all stocks on the exchange. In that case, you could get the result that the "mean" was much lower than the average, if there were a small number of stocks that were losing bucketloads of money.