This doctoral dissertation explores Finnish-speaking learners’ L2 pronunciation (second and foreign language pronunciation) of Finland-Swedish in free speech on different levels of oral proficiency. The aim is to show what kind of difficulties Finnish-speakers have with Finland-Swedish pronunciation and how pronunciation differs on different levels of oral proficiency, from A1 to C2 in the Common European Framework of Reference. In addition to the small number of studies in Finland-Swedish L2 pronunciation, previous research on L2 pronunciation in free speech is scarce even internationally, not to mention studies on pronunciation features on different proficiency levels. This dissertation comprises of a literature overview and three empirical studies. The literature overview maps out previous studies in different language pairs, including Finnish-speaking learners’ pronunciation of Finland-Swedish. The aim of the first study is to show what has and has not been studied regarding pronunciation in the chosen language pairs. The second study shows pronunciation of 9 grade pupils (n=68) at proficiency levels A1-B1. Their pronunciation is analyzed by four listeners with phonetic expertise. In the third study, the intonation of adult learners (n=8) at proficiency levels B1-B2 is studied acoustically by comparing the speakers’ fundamental frequency of speech (average f0, f0 variation, standard deviation, absolute mean slope, f0 peaks) and intonation contours with native speakers of Finland-Swedish (n=5). The fourth study focuses on advanced learners’ (n=12) pronunciation at proficiency levels B2-C2 as analyzed by Finnish-speaking and Finland-Swedish university students of Swedish (n=18). The results show that Finnish-speaking learners have native-like features on sound, word and sentence level in their pronunciation of Finland-Swedish. Learners also have difficulties with segmental and prosodic features despite the phonetic similarities between the languages. Quantity and word stress are mastered best even at the lower proficiency levels, whereas certain sounds, e.g. /ʉ/, /ʃ/ and /tʃ/, and prosodic features on sentence level, such as sentence stress and intonation, still cause difficulties for advanced learners. The dissertation shows that the overall rating for pronunciation improves as the proficiency level is higher. Improvement in sentence level prosody seems to have the strongest connection to overall pronunciation rating. Prosody is thus central in proficiency level assessment along with other features in assessment criteria. Based on the studies, mastering different features cannot, however, directly be combined with certain levels of oral proficiency as the learners had similar challenges at the different proficiency levels. On a more theoretical level, the results provide new information on pronunciation learning paths, which can be used in developing language teaching and assessment criteria.