STUDY QUESTION: Does the metabolomic profile, obtained with nuclear magnetic resonance (NMR), of spent culture media from human embryos correlate with reproductive potential in a cohort of good prognosis patients? SUMMARY ANSWER: In a large cohort of single transferred blastocysts from a homogeneous group of good prognosis patients, we find a high degree of individual variation in the metabolome that, however, has no relation to pregnancy outcome. WHAT IS KNOWN ALREADY: Differences among various specific metabolites have been linked to reproductive potential. Although results from retrospective near infrared (NIR) spectroscopy analyses of spent culture medias from transferred embryos were promising, randomized controlled trials were unable to demonstrate that NIR analysis improved pregnancy rates. Therefore, a more detailed investigation of the relation between embryo metabolism and reproductive potential is required. NMR is a powerful technique that provides detailed structural and dynamic information. STUDY DESIGN, SIZE, DURATION: A prospective cohort study was conducted at the Fertility Clinic, Aarhus University Hospital between February 2011 and July 2012. Infertile patients aged <38 years without endometriosis were offered participation and their embryos were included if greater than or equal to eight oocytes were retrieved. In total, 161 infertile patients were included in the cohort. PARTICIPANTS/MATERIALS, SETTING, METHODS: Spent culture media was collected on Days 3 and 5 after oocyte retrieval from 148 single transferred embryos. NMR spectra were obtained from 12 µl of spent media. Data were quantitatively analysed using multivariate analysis with respect to pregnancy outcome, defined as a live fetus by ultrasound in gestational Week 8, along with patient and treatment related variables such as embryo score, age, BMI, fertilization method and cause of infertility. MAIN RESULTS AND THE ROLE OF CHANCE: A total of 148 cycles were included in the analysis [embryo transfer cancelled (n = 12), no media collected (n = 1)]. Clinical pregnancy was confirmed in 47 patients (32%). We obtained high quality NMR spectra for 141 Day 3 and 137 Day 5 samples. Our spectra show a high degree of individual variation. Multivariate data analysis was performed on spectral data with several different pre-processing combinations, i.e. binning, alignment, normalization and scaling in the attempt to develop a valid prediction model. Different strategies of multivariate analysis showed, however, no correlation between the NMR profiles and pregnancy outcome, patient or treatment characteristics. No model could therefore be developed for prediction of pregnancy outcome. We conclude that within this group of good prognosis patients, large-scale metabolic variations between embryos detected with NMR have no apparent association with pregnancy outcome. LIMITATIONS, REASONS FOR CAUTION: Although this study is the largest we know of using NMR to investigate metabolomic profiles of single-transferred embryos, there may be differences that would be detected with a larger study. When analysing such a small sample volume, even small variations in the amount of media and dilution may introduce a large uncertainty in the results. WIDER IMPLICATIONS OF THE FINDINGS: Our study questions the usefulness of the entire metabolome for embryo selection, which should direct the search for viability markers in the culture media towards individual components. STUDY FUNDING/COMPETING INTERESTS: Funding was provided by Aarhus University, the Lippert Foundation, the Toyota Foundation, the Aase og Einar Danielsen foundation. Research at the Fertility Clinic, Aarhus Universtity Hospital is supported by an unrestricted grant from MSD and Ferring. The authors declare no competing interest. TRIAL REGISTRATION NUMBER: NCT01139268.