G*Power

I have used this over years to illustrate key statistical concepts to students; I plan to use it in a book on observer performance that I am working on; very intuitive interface;

This release marks a huge increase in the capability of G*Power, which now calculates for wide variety of designs. It is based on the notion of an effect size, and one of the useful features is its ability to calculate effect size from descriptive statistics and then transfer the result to the calculation of sample size. For many researchers, G*Power will be sufficient to see if your potential sample size will have a reasonable chance of capturing the effect you are looking for (assuming the effect exists!). The interface design makes calculations rapid and simple. The help files are online, and look well structured. Unfortunately, not all help topics are yet complete. Of course, if you don't know what sample size software should do, the application will make no sense to you!