The Kolmogorov-Smirnov (KS) test is a hypothesis test, which can be used
to test if a bunch of independent random variables drawn from the same
distribution is drawn from a given distribution or not. It can thus be
used to test if the CMB is Gaussian, but only if the CMB data have been
de-correlated before applying the test. I will discuss the application
of the KS test to CMB data, and show that the WMAP data are consistent
with the hypothesis of being Gaussian distributed with the powerspectrum
given by the best-fit Lambda-CDM model. I will then use the KS-test to
derive upper bounds on residual radio point sources in the CMB.
Potential applications of the KS test are the detection of Galactic and
extragalactic foregrounds, detection of primordial non-Gaussianity, and
cosmological parameter constraints.