I agree with pharreal that looking at players with <100 games is likely to lower the percentages as many would have abandoned poker before reaching 100 games.

While the proportion of players that make a profit is an interesting question. To me an even more interesting one is the proportion tournament registrations that are by players with a +ive ROI. Winning players are likely to play more often and multitable more often. So even though 27% of players on Stars are profitable a much higher proportion of registrations are likely to be by players with +ive ROIs. Here's some evidence to back this up.

A random sample of 10$ tournaments registering right now thanks to tournament selector.

9 tournaments on Stars with 45 players with 100 games, of those 45 33 have a +ROI and only 12 a -ive ROI. i.e. 73% of registrations are by winning players.

For Everest of 13 players registered 8 have +ROI i.e 62% winning players.

For Party Poker 37 players registered 19 of whom +ROI i.e. 51%

This is a pretty typical sample. So even though 1/3 of players might be profitable, 2/3 of players on the tables at the 10$ are likely to be profitable.

With tournament selector I reckon I can reduce that 2/3 of winning players down to 1/3 of winning players. THANK YOU sharkscope!

P.S. Sharkscope - Do you have any large datasets for tournament selector that could be datamined for % of winning players registering for tournaments of different sites. This would be a very useful tool especially if it could be split by buy level. Here's hoping.