This study investigates the differential effects of a task-based inside-school intervention and an outside-school project intervention on the development of speaking skills of 56 beginner second language (L2)-learners. The study extends previous research by looking at development across multiple occasions instead of two (i.e., a pretest and posttest), and by assessing speaking development by means of a non-intrusive task eliciting an informal conversation, rather than a formal test. To measure students' development of Dutch speaking skills, syntactic complexity, lexical diversity, accuracy, fluency, and communicative effectiveness were assessed. Multilevel analyses revealed differential effects for both interventions. The task-based intervention inside school led to significant positive development of communicative effectiveness and syntactic complexity measured by length and subordination, but also to more breakdowns in fluency. The outside-school project intervention, however, did not result in any significant speaking development. The findings demonstrate the learning potential of a task-based inside-school intervention for developing various aspects of informal speaking skills, except for fluency. From a methodological perspective, this study illustrates the complexity of test and measure selection when assessing L2 speaking development.