The molecular mechanisms that control the development of paired extremities are broadly conserved among vertebrate species. The paired fins of fish - pectoral and pelvic fins - are homologous to the fore- and hindlimbs of land vertebrates. Consequently, a fundamental knowledge about signalling processes in zebrafish paired fin development might help to understand limb patterning and congenital limb defects in humans. All-trans-retinoic acid (RA) is a key factor in many developmental processes including limb development. The current model for forelimb development was predominantly determined from studies in mice (Cunningham et al., 2013; Mic et al., 2002, 2004; Sandell et al., 2007; Zhao et al., 2009), chicken (Nishimoto et al., 2015) and zebrafish (Begemann et al., 2001; Gibert et al., 2006; Grandel & Brand, 2011; Grandel et al., 2002). It suggests an antagonism between RA and fibroblast growth factors (FGFs) along the anteroposterior axis, which mediates the correct positioning of the limb field and establishes a permissive environment for the induction of limb budding (Cunningham et al., 2013; Zhao et al., 2009). Moreover, RA cooperatively interacts with β-catenin signalling and Hox genes to control Tbx5 expression during forelimb development in chicks (Nishimoto et al., 2015). Examinations in zebrafish agree with the requirement of RA for pectoral fin induction (Gibert et al., 2006). For hindlimb development, however, the roles of RA are still controversial. The idea of a similar role for RA in fore- and hindlimb development (Nishimoto et al., 2015) contrasts with the opinion that RA is dispensable for hindlimb development (Zhao et al., 2009). In the zebrafish model, comparable studies investigating the role of RA on pelvic fin development are missing, which is why this thesis focused on this particular question. Gene expression analysis on zebrafish larvae revealed the presence of Rdh10a, Aldh1a2, Cyp26b1 and Cyp26c1 transcripts during the early stages of pelvic fin bud formation. The expression pattern of these genes, which are involved in RA synthesis and metabolism, indicated the establishment of an anteroposterior RA gradient in the early pelvic fin bud. Later, activity of RA signalling associated genes was detected along the forming fin rays. Based on heat-shock treatments of transgenic Hsp70l:Cyp26a1 zebrafish larvae, overexpression of Cyp26a1 and thus a reduction of the RA level was achieved during pelvic fin formation. From the obtained results an important role of RA in the development of pelvic fins during early stages of fin bud formation was concluded. A complete inhibition of the formation of endo- and exoskeletal pelvic fin structures could be achieved if the heat-shock treatment was started before the first signs of a morphological fin bud appeared. After the onset of fin bud formation, Cyp26a1 overexpression resulted in the reduction of the overall length of the pelvic girdle accompanied by the lack of diverse skeletal elements, mostly the posterior process and the radials. These results indicate a putative role of RA in the pelvic fin initiation process, which seems to occur during a limited time frame. Moreover, they suggest a role of RA in pelvic girdle patterning and chondrogenesis. Additionally, a participation in fin ray formation and growth is likely. However, since the entire organism is affected in these experiments, unspecific effects cannot be ruled out. Therefore, the main focus of this work was to establish the binary Gal4-UAS system with the aim to manipulate RA signalling in a spatially and temporally controlled manner. On the one hand, driver lines provide the expression of either a hormone- or light-inducible Gal4 variant under the control of tissue-specific enhancers. Here, three Gal4 variations - ERT2-Gal4-VP16, KalTA4-ERT2 and GAVPO (Akerberg et al., 2014; Distel et al., 2009; Gerety et al., 2013; Kajita et al., 2014; Wang et al., 2012) - were investigated and considered suitable for the use in zebrafish. Tissue-specifity was achieved by selecting enhancers of the genes Prrx1a, Prrx1b and Pitx1, which are active specifically in pectoral and/or pelvic fins (Chan et al., 2010; Hernández-Vega & Minguillón, 2011). On the other hand, effector lines express genes encoding either a dominant-negative retinoic acid receptor (dnRarα2a) (Stafford et al., 2006) or the RA metabolizing enzyme Cyp26a1 under the control of five repetitive (5x) or four non-repetitive (4xnr) upstream activating sequences (UAS) (Akitake et al., 2011; Goll et al., 2009). Driver and effector constructs are equipped with minimal Tol2 cis sequences mediating transgene integration into the genome by Tol2 transposase activity. Moreover, different marker genes facilitate the identification of single or multiple transgenic zebrafish. As a proof-of-principle, the activation of dnRarα2a expression in F3 embryos of 5xUAS:dnRarα2a-IRES-eGFP zebrafish by injection of KalTA4-ERT2-GI mRNA, followed by induction with 4-hydroxy-tamoxifen (4-OHT) was demonstrated. Altogether, the basis for a valuable genetic tool was created, that combines several advantages: a simple and practical application, a simplified screening process, the visualisation of transgene activity and the optimization for the zebrafish model organism.