Estimates of air infiltration in houses based on tracer gas measurements have usually assumed house is a single perfect mixing chamber with incoming air instantaneously and uniformly diffused to all parts of the interior. Points out that in reality some parts of the house - basement or rooms with doors closed - exchange air only very slowly with other parts so that actual mixing is far from instantaneous. Presents theory and mathematics necessary to apply tracer gas method to buildings of many chambers. NOTES A letter commenting on the use of the term "diffusion" with reply by Sinden is published as "Letters to the editor:multi-chamber theory of air infiltration" M.Sandberg. Build. Environ. vol 14 no 1 1979 pps 69,70.