This thesis investigates advanced Swedish students’ development of three grammatical phenomena: subject-verb concord, prepositions and article use in compositions and translations. In order to describe the students’ development of these categories, actual errors are related to potential errors forming so called ‘error scores’. When the error score for the first test opportunity is compared with the last one, three different developmental patterns appear: a progression pattern, a regression pattern and a U-curve developmental pattern. In addition, in each category there are students who are error-free, and who thus form a zero group. As for the students’ mastery of subject-verb concord, errors with subject noun heads that contrast as to number in Swedish and English are found to induce a considerable number of errors. It is also found that it is with these subject heads that students progress the most. However, since such errors are taken to be lexical in nature, they are considered separately from ‘true’ concord errors. Once errors involving contrastive lexical heads are removed, students appear to fossilise with subject-verb concord in both compositions and translations, although it was found that students experience considerably and consistently more problems with noncontiguous subject-verb constructions than with contiguous ones. The students’ development of prepositional uses is considered from a predictability continuum. ‘Basic’ prepositional uses are defined as uses that are easy to predict because they are identical in the two languages. ‘Systematic’ prepositional uses are believed to be somewhat more difficult to predict than ‘basic’ prepositional uses since they involve the memorising of rules and prefabricated patterns. (It is the introduction of this prepositional category that justifies the inclusion of prepositions in a study of grammatical phenomena.) ‘Idiomatic’ prepositional uses are hypothesised to be most difficult to predict since they involve item-by-item memorisation and cannot be generalised in the way ‘systematic’ prepositional uses can. The hypotheses are confirmed in several ways. For example, the error score is lowest and the zero group largest with ‘basic’ prepositional uses. With ‘systematic’ prepositional uses, the error score is higher and the zero group smaller than for ‘basic’ prepositional uses. The highest error score and the smallest error-free student group appear with ‘idiomatic’ prepositional uses. The students’ mastery of article use focuses on the generic use of the zero article (noncount nouns and plural nouns) and the use/non-use of the definite article before proper nouns. In both cases, high average error scores appear at all test opportunities. As for the generic use of the zero article, it is suggested that advanced students here experience major problems because, whereas there is a link between the indefinite form and genericness in English, there is no such firm association in Swedish where both forms, depending on different variables, can be used to express genericness. This may cause the very concept of genericness to become fuzzy for Swedish learners of English. As for the use/non-use of the definite article before proper nouns, the suggestion is put forward that students, due to the great number of rules in this area, resort to memorising each individual proper noun. This would explain why there is no discernible progression in average error scores in this category, which would be in line with the finding that constructions that involve non-generalisable aspects of grammar, such as subject-verb concord with contrastive lexical heads and ‘idiomatic’ uses of prepositions, are the most difficult ones to master for advanced Swedish learners of English.