Increases in atmospheric greenhouse gases may change the hydrology of a num
ber of Canada's regions. This will have an impact on aquatic and wetland ec
osystems as well as municipal, industrial, and power generation uses. It is
thus important to get an estimate of the potential changes to the Canadian
hydrological cycle in order to make intelligent decisions concerning mitig
ation factors that society may be forced to undertake. We divided Canada in
to ecoclimatologically similar regions called "ecozones." We developed two
month-stepped temperature-precipitation-runoff models for the country using
an artificial intelligence neural network (ANN) approach. We modified inpu
t temperature and precipitation variables in the ANN models to match those
predicted by the Canadian Climate Centre General Circulation Model II for a
doubled CO:! atmosphere and calculated new monthly equilibrium runoff pred
ictions. Our results predict that much of Canada will experience higher ann
ual runoff than is currently the case. The timing of runoff will change sig
nificantly in a number of the ecozones, as we show that in many regions, pe
ak runoff will occur approximately 1 month earlier than is currently the ca
se. The ANN model did not work as well for basins in the Prairie ecozone, a
s we could not develop a good model with data from regulated rivers.