This article examines the predictors of second-language proficiency for a group that until now has hardly been investigated: immigrants who rarely participate in the host society and who have a low level of second language proficiency (sample characteristics are for example: no paid job, low educational and literacy level, high mean age and number of years since migration). In contrast with earlier research, not only self-assessments were used as indicator for second-language proficiency, but also language test scores. Results from a sample of 624 immigrants partly replicate findings from earlier studies: self-assessed second-language proficiency is higher among immigrants who have followed a language course, do voluntary work, have a high educational level, high mother-tongue proficiency, a low migration age, and a large number of years since migration. No links, however, were found between having psychological problems, gender, and migration motive and self-assessed second-language proficiency. Furthermore, some new predictors of self-assessed second-language proficiency were identified,namely similarity in alphabet between mother tongue and second language, daily interactions with natives in the public domain, and speaking the second language at home. For a subsample (N¼98) second-language proficiency was also assessed via (objective) lexicon tests. When using scores on this test as dependent variables, only years since migration turned out to be a significant predictor. Though certainly tentative, this finding indicates that different predictors of second-language proficiency may apply depending on how it has been measured.