The reference entry contains two datasets suitable for benchmarking feature-finding/quantification algorithms for LC-MS-based metabolomics. The first dataset (MTBL234) contains a spike-in experiment of selected compounds into a human plasma metabolome background. Seven different compounds (four of them with stable isotope labels) were added in a series of 10 concentrations covering four orders of magnitude. Data was acquired on a Q-TOF mass spectrometer (Waters Aquity ultra-high-performance LC system coupled to a Synapt HDMS oa-Q-TOF mass spectrometer). The second dataset (MTBL235) consists of a synthetic data set based on published plant metabolite study. The data contains simulated profile MS spectra of an LC-MS experiment based on hundreds of identified compositions and experimentally determined retention times and thus gives an accurate example of a (low complexity) metabolome dataset. In contrast to real-world datasets, a known ground truth is available and thus precision and recall of the feature finding algorithms can be assessed. Details of the datasets can be found in the methods section of Kenar et al. (PMID 24176773), together with details of the performance.