Natural products are of great importance and value in many industries. However, the ability to produce these compounds limits the usages and hampers development of e.g. new pharmaceuticals. Glucosinolates are amino acids-derived specialised metabolites, which have been shown to improve health in humans and therefore have gained interest as potential nutraceuticals. Accumulation of glucosinolates is limited and strictly regulated in native producers. This regulation makes increasing de novo synthesis difficult. Establishment of production in alternative organisms have been attempted. However, the biosynthetic pathways contain many of the complexities often found in plant pathways, which make them difficult to express efficiently in heterologous hosts. Therefore, the titres have so far remained low. In this thesis, we tested three different host systems for their ability to produce glucosinolates. The native producer Arabidopsis thaliana, and two microbial hosts Escherichia coli and Chlamydomonas reinhardtii. Some glucosinolates are derived from chain-elongated amino acids and thus both chain elongation and core structure pathways were used in constructing heterologous production hosts. Pathway engineering attempts in C. reinhardtii were unsuccessful, but based on a thorough review of literature suggestions were made to facilitate future engineering endeavours in this host.Previously reported problems of by-product formation when expressing the chain elongation pathway in E. coli was tackled by construct and protein engineering approaches. This work revealed that methylthioalkylmalate synthase (MAM) 1 instead of MAM3 was responsible for in vivo chain elongation of phenylalanine for the synthesis of 2-phenylethyl glucosinolate. The by-product formation could be removed through medium optimisation. Moreover, production of benzyl glucosinolate was established in the bacterial host E. coli and optimised through construct design, cultivation conditions and strain engineering to produce levels similar to total glucosinolate content of wildtype A. thaliana plants. Finally, multi-gene mutant lines of A. thaliana Col-0 were generated, in which glucosinolate production was targeted towards 4-methylsulfinylbutyl glucosinolate production. These lines exhibited notable developmental phenotypes and tissue-specific changes in glucosinolate profiles.