Second language (L2) writing (especially English as a foreign language) is generally considered as one of the most important skills that language learners need to acquire during their language education (Lee, 2016). However, L2 writing might be one of the most difficult skills for English as a foreign language (EFL) learners to develop throughout their language learning process (Barkaoui, 2007; Frydrychova Klimova, 2014). Consequently, writing has often been found to be one of the weakest skills of EFL learners around the world (“IELTS Test taker performance 2016”, 2016), including among Hungarian language learners (Csapó & Nikolov, 2001, 2009; Mihaljević Djigunović, Nikolov, & Ottó, 2008). In Hungary, as in other contexts, although writing might be one of the most difficult skills to acquire, language learners have been found to spend an insufficient amount of time learning to write in the foreign language classroom (Árva, 2007; Cook, 2005, 2008; Nikolov, 2002). In addition, research has shown that Hungarian university students make limited use of self-regulatory strategies during writing, despite the important role they play in second language writing development (Kormos, 2012; Lam, 2015; Nitta & Baba, 2015, 2018). Research on L2 writing development in the Hungarian context is therefore important to establish where particular challenges may exist, and how these might be addressed in language education programs. L2 writing development has been mainly investigated by studies adopting a two-wave longitudinal research design (Barkaoui, 2016; Bulté & Housen, 2014; Knoch, Rouhshad, & Storch, 2014; Knoch, Rouhshad, Oon, & Storch, 2015; Mazgutova & Kormos, 2015; Storch, 2009; Storch & Tapper, 2009). In a two-wave longitudinal research design, changes in complexity and accuracy indices are measured at two points in time from a large population. Although two-wave longitudinal studies are useful in pointing to tendencies and making generalisations in L2 writing development, most recent research suggests that L2 writing development is idiosyncratic, that is, no two learners exhibit similar developmental patterns (Chan, 2015; Rosmawati, 2016). The 2000s saw the emergence of an influential approach to second language acquisition, the Complex Dynamic Systems Theory (CDST), which acknowledges that there is no average language learner (de Bot, Lowie, & Verspoor, 2007). According to the CDST, language is seen “as a dynamic, complex and nonlinear process” (Larsen-Freeman, 1997, p. 142). CDST studies usually adopt a multi-wave research design to gain insight into the dynamic and complex nature of language development (Caspi, 2010; Larsen-Freeman, 2006; Spoelman & Verspoor, 2010; Verspoor, de Bot, & Lowie, 2004; Verspoor, Lowie, & van Dijk, 2008; Verspoor, Lowie, Chan, & Vahtrick, 2017; Verspoor, Lowie, & Wieling, 2018). This study investigates the second language writing development of four Hungarian EFL learners over a nine-month period by adopting a multi-wave mixed-methods research design. The participants were enrolled in an EAP programme offered by a university in Budapest, Hungary. Two argumentative essays were composed each month by the four participants in their naturalistic setting. Furthermore, one argumentative essay was written by the four learners on a monthly basis under a controlled setting. The final mini corpus consisted of 92 argumentative essays collected from the four participants over the nine month period. The four participants were also interviewed on their self-regulatory processes over the nine-month investigation after each controlled written sample was collected. This study found that lexical and syntactic complexity indices developed nonlinearly in the four participants’ written data over the nine-month investigation, substantiating previous studies on L2 writing development (Caspi, 2010; Larsen-Freeman, 2006; Rosmawati, 2016; Spoealman & Verspoor, 2010; Verspoor et al., 2004, 2008, 2017, 2018). However, the directions of the trends were different for all lexical and syntactic complexity and accuracy indices in the four EFL learners’ written data. Lexical and syntactic complexity and accuracy indices showed a great deal of variability in the four learners’ written data over the nine month period. The amount of variability constantly changed in the complexity and accuracy indices over time. In addition, the degree of variability was also different in the four participants’ written data which supports previous studies on L2 writing development (Caspi, 2010; Larsen-Freeman, 2006; van Dijk, Verspoor & Lowie, 2011). This study found that there was a general improvement in one of the participants’ lexicon and in another participant’s accuracy over time as demonstrated by the statistically significant developmental peaks (van Geert & van Dijk, 2002). The interactions between lexical and syntactic complexity and accuracy were dynamic over the nine-month investigation. The polarity of the interactions between lexical and syntactic complexity and accuracy changed from negative to positive and vice versa over time. Moreover, the magnitude of the interactions oscillated over time, ranging from weak to strong associations. The polarity and the magnitude of the interactions within lexical and syntactic complexity and between lexical and syntactic complexity and accuracy were different in the four participants’ written data which supports previous studies on L2 writing development (Spoelman & Verspoor, 2010; Verspoor & van Dijk, 2011). The self-regulatory processes developed nonlinearly and at different rates in the four participants’ learning journey. The four participants’ focus shifted from self-observation to self-evaluation processes at different points in time which substantiates earlier studies on self-regulation (Nitta & Baba, 2015; Sasaki, Mizumoto & Murakami, 2018). In addition, only two learners’ focus shifted from self-evaluation to goal-setting over the nine-month investigation. The findings of this study bear important theoretical, methodological and pedagogical significance and implications. First, the findings of this study support the Complex Dynamic Systems Theory perspective on second language development. Second, this study demonstrated how quantitative and qualitative CDST methods can be combined. Third, this study presented pedagogical implications for teaching and language learning. By demonstrating that language development is dynamic and idiosyncratic and language development is the outcome of the self-organisation of the language systems, this study questions the validity and the effectiveness of an item-processing approach. This study offers a new approach to the study of second language writing development and advances our understanding of the complex and dynamic nature of second language development itself.