Over the past few years the interest in metabolomics has increased in various fields including forensic toxicology. Forensic analysis typically requires a high degree of accuracy, which is often a problem in metabolomics applications. We aimed for a systematic evaluation of different analytical considerations of a metabolomics workflow allowing a targeted approach within an untargeted setup. Samples with 69 metabolites from different chemical classes were qualitatively and quantitatively analyzed on a high resolution quadrupole time of flight mass spectrometer coupled to liquid chromatography (UHPLC-QTOF). Three different questions should be addressed: (1) two different approaches on 'blind matrix', a simulated body fluid (SBF) and plasma-filtrate were tested for calibration samples, (2) comparison of two different HPLC-columns, reverse-phase (RP) and hydrophilic interaction chromatography (HILIC), and (3) comparison of three different acquisition modes (TOF-MS, information dependent data acquisition (IDA) and sequential window acquisition of all theoretical fragment-ion spectra (SWATHSamples were measured repeatedly for method comparison based on sensitivity, accuracy, precision and detection robustness. The blind matrices showed similar accuracy for most analytes, while SBF provides an easier preparation with satisfying results. To cover a wide part of the human metabolome, a combination of RP and HILIC showed best results. The different scan modes performed equally regarding metabolite quantification while TOF-MS was more sensitive but lacks of MS/MS spectra generation. IDA and SWATH files were aligned to various databases where IDA showed good MS/MS spectra matches. SWATH seems to be beneficial in detection rate but is incompatible with many important software tools in metabolomics.