Problem is that nobody has even a rough estimate. 300k, for example, seems way too high for "direct civilian deaths" in the middle-east if the term means "killed by U.S. Delivered munitions." The iraqbodycount.org site shows the figure to be under ten thousand for deaths caused by "US Led coalition", and almost all of those in the first month.

There have been a number of efforts, some quite controversial, on the effort to find the tally of indirect deaths in Iraq (I.e. How many people would not have died had there not been a war on at all), some of those figures approach or in one or two cases well exceed that 300k figure.

Iraqbodycount is an interesting case, as it relies on substantiated reports resulting from the fact that there is a competent news agency continuing to operate, with reasonable communication. Places like Afghanistan get much more difficult.

A further problem is that usually people aren't all that worried about just who killed who, as much as there was a violent death. As a result, you will hear figures thrown about for "civilians killed in Afghanistan", sometimes it will say "by Afghan and ISAF Forces", but rarely will it break things down further. Figures for the initial month or two (about 5,000 in the case of Afghanistan) are usually simpler to understand as the invading forces are usuall the ones doing the killing. When you get into a situation of additional indigenous forces starting to stand up and partake, and particularly in the case of Iraq, when you get numbers of independent groups in internecine warfare, figures become very difficult to distinguish.