I use R (www.r-project.org/) and am generally happy with its plots. The graphing commands have a few quirks, and a lot of non-orthogonal inputs, but it lets yo

I use R (www.r-project.org/) and am generally happy with its plots. The graphing commands have a few quirks, and a lot of non-orthogonal inputs, but it lets you dig in and do things at a pretty low level, in case you want to fine-tune your graphs. One thing I remember fondly from Matlab days, that R isn't so good at, was the ability to grab a handle to a graph axis, and rescale the axis, without starting a new plot. R pretty much forces you to know, from the start, what limits each axis should have. A colleague uses SciPy and is similarly very happy. However, I can't speak about it from personal experience. I've used gnuplot (www.gnuplot.info/) for small tasks (one or two lines on an x-y plot). My general impression is I would find it too constricting for bigger jobs. In particular, R allows me to do a full-fledged analysis and then plot results, while gnuplot pretty much requires you to generate the results externally.