I agree with Eduard - but make sure you use something like the ggplot add-in. The nice thing about R is that you can export the graphic in the format and size the journal requests - no need to fiddle around in post-production.
Do you know the program called Eviews? It´s an econometric software and very useful and easy. The software has a lot of graphics for different uses. For information: http://www.eviews.com/home.html
R is the most flexible, has packages for nearly everything and is free to boot! It has a bit of a learning curve however. Also, the people who maintain it have very strong opinions about what constitutes an informative and accurate graph. Some graph types are actively discouraged by a complete lack of implementation. A bargraph with error bars will take about half a page of script, for instance, whereas a boxplot is just one short command. Takes some getting used to :-).
If you're in the biomedical field you could also have a look at GraphPad Prism. It's one of the easiest programs to use for basic statistics (almost too easy, takes the fun right out of it) and makes very pretty graphs. It's not free, but a whole lot cheaper than SPSS. Only downside is that you can't do any multivariate statistics as it uses a broad form spreadsheet layout.
As Hendrik said R gives you a lot of room to customize your graphs. In general you get a better insight of what stands behind the test/operation you just done. As also mentioned above with the ggplot package you can build really nice graphs. Please look at these websites for tutorials/documentation:
SigmaPlot has been my choice for a long time as it's extremely flexible when it comes to design details. The drawbacks have been a fairly hefty price and frequent "updates" which always cost you, but come with very slight improvements. The current version is 12, but I'm happily using version 8 as I'm not aware of any MAJOR improvements in the basic graphics designs I use.
ggplot or lattice in R. Python also support graphics. Sometimes i prefer to use python instead of R. Here is an example of seaborn, a pyhon graphic tool;