Untargeted liquid chromatography-mass spectrometry (LC-MS)-based metabolomics studies are usually carried out in both positive and negative ion modes; however, it is frequently ignored that the optimal conditions in positive ion mode and negative ion mode are often not the same. We carried out a systematic investigation on urine samples to evaluate the additive effects in negative ion mode. It was found that the widely used conditions, 0.1% formic acid (FA) and NH(4)Ac at different pH, are far from the optimum for untargeted urine metabolomics studies. Compared to 0.1% FA, the use of 1 mM acetic acid (HAc) resulted in almost three times as many detected peaks (401 vs 148) and around five times the size of the peak area (33.55 × 10(6) vs 6.47 × 10(6)). The remarkable improvement can be explained by two factors: (i) a significantly enhanced ionization efficiency due to the combination of an appropriate pH at around 4.0-4.5, the reducibility of H(+), and the high gas-phase basicity of Ac(-) and (ii) a reproducible LC separation due to an acceptable buffering capacity. Our study revealed the importance and necessity of additive optimization, which can be of benefit in related metabolomics studies.