This systematic review aims to explore how task complexity and task sequencing affect the complexity, accuracy, and fluency (CAF) of second language performance. Through an analysis of studies conducted between 2012 and 2021, this review investigates the characteristics of the studies, operationalisation of task complexity, measurements of CAF, and the effects of task complexity and task sequencing on CAF. The results suggest a shift in research focus from task complexity to task sequencing. Most prior studies examined Robinson's hypotheses among English adult learners engaged in writing tasks, mainly utilising elements, reasoning demands, and content support to manipulate task complexity. In addition, they employed a diverse range of methods to measure CAF. The findings partially support the Cognition Hypothesis and the SSARC (stabilise, simplify, automatise, reconstruct, and complexify) model. Previous findings provide robust evidence that complex tasks, controlled via elements and reasoning demands, enhance language complexity. Furthermore, sequencing tasks from simple to complex has been shown to overall improve CAF. Implications for future research are discussed.