In an earlier paper, I outlined the methodology for the demographic estimation of language shift (de Vries 1974). The method was applied to the data in Finland for the intercensal decade 1951-1960. A similar method was applied in an analysis of language maintenance among Ukrainians in Canada (de Vries 1977, forthcoming). In the present paper, the same method is used again, with minor modifications, to estimate the net effects of language shift on the Swedish-speaking population of Finland for the decade 1961-1970. Briefly, the estimation technique is a 'residual cohort survival method' in which we use initial and final frequencies for birth cohorts by sex and main language, estimate intercensal mortality and net migration for all age-sex language cohorts, to arrive at estimated frequencies by cohort. Any discrepancies between observed and estimated frequencies can then be attributed to one or more of the following: (a) data errors; (b) deviations from basic assumptions underlying the estimation, and (c) residual processes, in this case language shift. For a more detailed exposition of the method, see de Vries 1974, or the references cited there.