We developed a method for determining whether channels in a multichann
el patch or bilayer have the same or statistically significantly diffe
rent open probabilities. We use a maximum likelihood method to fit the
distribution of (unbinned) current amplitudes and to provide estimate
s of individual channel open probabilities, single channel currents, a
nd standard deviations of the channel currents. These parameters are u
sed to compare models with increasing constraints on the open probabil
ities including the model where all channels have different open proba
bilities and the model where all channels have the same open probabili
ty. A chi(2) statistic is used to identify models that are statistical
ly less likely to predict the data. The ability of multichannel data t
o determine individual open probabilities is limited by two factors: t
he signal to noise ratio of the record and the fact that changes in am
plitude distributions caused by a 0.2 difference in open probabilities
are comparable in magnitude to the variations caused by random channe
l gating. These limitations notwithstanding, we demonstrate the utilit
y of our approach by using it to analyze the open probabilities of 3 l
arge conductance Ca2+-activated K+ channels in an artificial lipid bil
ayer revealing the response of one of those channels to GTP gamma S.