Much of recent literature on bilingualism and cognitive development has claimed that bilingualism causes cognitive enhancement. This paper reviews the methodology of these studies and concludes that these, primarily associational', studies do not support this conclusion. A more parsimonious explanation of differences between bilinguals and unilinguals is to be found in differences in self-selection of children in non-compulsory second language learning environments and in difficulties of access to learning and social stress which affect minority language children who are forced to learn the majority language in ways that are very much like the difficulties of lower class children where language is not an issue.The 1962 Peal and Lambert study seems to have marked a watershed in research into the relation between bilingualism and various cognitive skills (Cummins, 1976b(Cummins, , 1978. Before that, the weight of research evidence tended to support a negative association. That is, comparisons of groups of unilingual and bilingual children on various verbal and nonverbal intelligence measures indicated some superiority for the unilinguals more often than not. Although some current research still supports the negative association, the trend of the evidence has swung to favour the bilingual children. Among research done in a Canadian setting, I have found no recent work which puts unilinguals ahead of or equal to their bilingual peers.This paper attempts to account for this shift in results and to examine the claim that there is a 'causal link between bilingualism and (cognitive) flexibility, the former apparently enhancing the latter' (Lambert, 1978: 539). The review of the literature will consider first the research designs and the results. Next, the dependent, cognitive measures will be discussed and then sample selection and the operationalization of Linguistics 17 (1979), 231-255.