Urine-based metabolomics-driven strategies for the discovery of biomarkers are increasingly developed and applied in analytical chemistry. But valid, data-based recommendations for a urine sample material of choice are lacking. We investigated first and second morning urine (MU), which are the most commonly used urine specimens. Potential major factors biasing metabolomics biomarker results in these sample materials were studied. First, 35 1st and 2nd MU samples were collected from healthy, young men after an overnight fast. Subsequently, two subgroups were built, one having fast food at lunch and dinner (n = 17), the other vegetarian meals (n = 18). Again 1st and 2nd MU were collected. Non-targeted liquid chromatography-mass spectrometry was applied for analyses. More than half of the >5400 urinary ion features showed a significant difference between 1st and 2nd MU. Just two fast food meals on previous day significantly affected around 30% of all metabolites in 1st and 2nd MU. In contrast, the effects of two vegetarian meals in 2nd MU were only minor. Additionally, we describe 47 metabolites in urine, possible hits in biomarker studies, which are susceptible to the diet the day before sample collection. They should be handled with caution until validation in diet-controlled studies. Based on our results we think the second MU, ideally collected after standardized vegetarian meals and drinking only water on the previous day, is most suitable for valid analysis of biomarkers in urine.