Most of the research on the assessment of the intelligence of Latinos in the United States appears to be based on some possibly erroneous or at least dubious assumptions. Among these are the following: (1) the assumption of bilinguality; (2) the assumption of equal proficiency in the English language; (3) the assumption of the equivalence of literal translations from English to Spanish; (4) the assumption of population homogeneity with respect to socioeconomic, religious, and ethnic factors; (5) the assumption of the language uniformity of Spanish; (6) the assumption that testing can be culture-free; (7) the assumption of the superior quality of investigations of Latinos conducted by Latinos. This paper evaluates the evidence for each of the above assumptions, and concludes that great caution should be exercised to avoid taking them too much for granted. (Author)