MATLAB

hermquad.m
Calculate the Gauss-Hermite abscissae and weights, given the number of points you wish to use.

invpsi.m
Calculate the inverse psi (digamma) function.

pinwheel.m
Generate two-dimensional noisy pinwheel data.

warpmesh.m
This applies a matrix to every entry generated by a call to meshgrid().

zerocross.m
A very simple function to calculate the locations of zero crossings in a vector.

Python

xmpl
This lets you use Matplotlib (a very nice graphing library) to generate plots without having to actually write any Python. The idea is to make an XML document that is then rendered.

Perl

Template::XMPL
This is for using xmpl (see above) with the Template Toolkit. With this you can use Perl and Catalyst (or whatever) on your webserver to generate plots from (perhaps dynamically-generated) XML.

Lisp

cl-gnuplot
A basic package for manipulating Gnuplot from Lisp.

Java

XYZtoMorpheus
A little program for converting between morphometric formats.

Links to Other People's Useful Code

Lightspeed
Tom Minka's incredibly-useful MATLAB toolbox. It introduces many useful functions and speeds up many existing ones.

Gaussian Processes for Machine Learning
The MATLAB code to go with the book by Carl Rasmussen and Chris Williams.

Minimize
A great conjugate gradient minimizer for MATLAB by Carl Rasmussen.

mmfminunc.m
MATLAB Quasi-Newton optimizer. Very nice when you don't have an analytic gradient. Written by Duane Hanselman.

scattercloud.m
Plots a histogram of two dimensional data in MATLAB. Written by Steve Simon.

Stixbox
If you don't have the MATLAB Statistics Toolbox ($$$), this might not be a bad substitute.