This thesis investigates the command 8th-graders in Arabic home language instruction have of written Modern Standard Arabic and if the type of instruction they have received and/or contact with written Arabic affect their performance. Background chapters discuss variables connected to the Arabic language (diglossia, research on reading and writing in Arabic) and variables connected to HLI in Sweden (set-up, steering documents).  The testing material consisted of a translation test from Swedish to Arabic combined with a questionnaire that addressed various factors of relevance to language acquisition.  The translations were analysed on three levels: (1) handwriting, (2) spelling and (3) morphosyntax. The main result of the analysis was that the participants were highly heterogeneous: some participants produced incomplete translations in handwriting that was barely legible, whereas others had good results for all measures. Many of the participants relied on a phonological strategy for spelling. For example, even short, high-frequency words such as personal pronouns and prepositions had not been spelled correctly.  The results for handwriting, spelling and morphosyntax were checked against the variables (1) years of HLI, (2) extra instruction in Arabic outside of HLI and (3) contact with written Arabic in the free time. The results for the effect of participation in HLI were inconclusive. However, many, but not all, of the participants with good results on the translation test had received extra instruction in Arabic, either in Sweden or prior to coming to Sweden. Reading Arabic in the free time was not in all cases connected to good results, but not reading Arabic in the free time was in most cases connected to a low command of written Arabic. Regarding these results, it is suggested that additional factors (motivation, support from the family, etc.) could be at play.  Previous research has addressed the question of heterogeneity in HLI classes. The findings of this thesis illustrate how great the heterogeneity can in fact be, and thus have implications for the set-up of Arabic HLI in Sweden.