R (http://cran.r-project.org/) together with  RStudio (http://www.rstudio.com/). There is no better choice.

R (http://cran.r-project.org/) together with  RStudio (http://www.rstudio.com/). There is no better choice.

have you ever experienced the open source scilab tool ?

http://www.scilab.org/

I use SPPS, Systat ,  you can try SAS , but it depends on the type of work you do

Just to add more depth to my previous answer, why is R/RStudio an excellent choice to perform statistics on the Mac (and also on Linux and Windows, for what matters)?

The pros:

- Free and open source: you can install and use it everywhere, without licensing issues;

- Huge number of packages for extending the base application and for handling specialized/specific tasks. Actually, the number of packages is so huge that one needs specific search tools to find what could be useful, e.g.,  http://cran.r-project.org/web/views/, http://dirk.eddelbuettel.com/cranberries/, http://www.rseek.org/;

- Excellent support: several websites post introductory and advanced articles and tutorials about R and its packages (http://www.statmethods.net/, http://www.r-bloggers.com/, http://learnr.wordpress.com/, to name a few) and you can also find several interesting papers on R-Journal (http://journal.r-project.org/index.html), an online magazine devoted to R;

- Easily programmable: if you have some basic programming skills you can easily write scripts to automate your statistical analyses.

Cons:

- Steep learning curve, but RStudio eases things considerably. And that is science, after all.

It really depends on whether you are comfortable with a command line approach of stats  in which case I could recommend R, which is free and flexible plus one can interact with other users and share their experiences. The other platforms would include stata and, spss

Hi

SPSS or XLSTAT

JMP7 is also good