A mark-recapture experiment was conducted applying a two-sample strati
fied technique to estimate the number of Atlantic salmon smelts, Salmo
salar, migrating out of the Conne River, Newfoundland. We developed a
model where parameters are introduced to describe the mean time for t
he salmon to migrate between the release site and recapture site and t
o describe the probability of capture on a particular date. The latter
are then used to expand the number of untagged smelts captured to est
imate the daily run sizes. We discuss the advantages of this new appro
ach over other models. For example, this formulation allows great flex
ibility in the experimental design, is robust to violations of assumpt
ions, and allows the influence of environmental factors upon the param
eters to be investigated.