Insects are one of the most abundant and diverse animal groups, and they include many valuable ecological indicator species, but taxonomic discovery projects and biodiversity surveys targeting this group are often challenging. While mass trapping devices allow the collection of insects in great numbers, the task of identifying the species present is a painstaking and resource-demanding process. Metabarcoding, that is, high throughput sequencing of PCR-amplified species-specific genetic markers in environmental samples, promises to solve this problem. However, metabarcoding is still in its infancy. In this thesis, I optimized metabarcoding methods for inventorying and accelerating species discovery of terrestrial insects. In paper I, we designed new PCR primers for mitochondrial markers and evaluated them against existing ones using in silico methods. We showed that the best marker for metabarcoding of insects is 16S because of its broad taxonomic coverage and low amplification bias. However, there is significantly more reference data for COI, and its taxonomic coverage is reasonable when using sufficiently degenerate primers (mixes of primer sequences). In paper II, we applied 16S and COI metabarcoding to different types of samples of the same insect communities: Malaise trap samples (preservative ethanol or homogenized samples) and soil samples. The results show that the two-marker strategy increases biodiversity detection over single-marker analyses. They also show that 16S is better than COI for metabarcoding of eDNA samples because the less degenerate 16S primers do not amplify as many off-target organisms. Finally, the results show that analyses of tissue homogenate and preservative ethanol yield strikingly different results. Large and heavily sclerotized insects do not leak DNA into preservative ethanol like small and weakly sclerotized ones do, but their DNA tends to swamp the DNA of the latter in homogenized samples.  In paper III we evaluated the performance of various non-destructive mild lysis treatments and DNA purification methods. We subjected mock community samples to incubation in either a milder or a more aggressive digestion buffer for a short or a long period of incubation. The DNA was then extracted using either a manual or an automated purification protocol. We found that the milder digestion buffer and the shorter incubation time preserved the morphology of the insect best while at the same time giving the most accurate DNA metabarcoding results; the purification protocol had little or no effect on metabarcoding results. Finally, in paper IV, we explored the received wisdom that high concentrations of ethanol, although optimal for preservation of the DNA, make insects fragile and difficult to work with from a morphological point of view. We preserved insects in different ethanol concentrations and subjected them to damaging processes, such as shaking or transporting. We verified that high concentrations of ethanol induce brittleness, although the effect is less pronounced in robust insects. Our results also indicate that shipping by mail is safe for samples preserved at intermediate concentrations (70 or 80 %). In summary, this thesis represents a significant step forward in the development of methods for preserving and analyzing samples of terrestrial insects for biodiversity surveys, monitoring programs, and taxonomic research projects.