OBJECTIVE Magnetic resonance spectroscopy (MRS) measures hepatic fat and adenosine triphosphate (ATP), but magnetic resonance studies are challenging in obese subjects. We aimed to evaluate the inter- and intrarater reliability and stability of hepatic fat and ATP measurements in a cohort of overweight and obese adults. METHODS We measured hepatic fat and ATP using proton MRS ((1)H MRS) and phosphorus MRS ((31)P MRS) at baseline in adults enrolled in the Action for Health in Diabetes (Look AHEAD) clinical trial at one site. Using logistic regression, we determined factors associated with successful MRS data acquisition. We calculated the intra- and inter-rater reliability for hepatic fat and ATP based on 20 scans analysed twice by two readers. We also calculated the stability of these measures three times on five healthy volunteers. RESULTS Of 244 participants recruited into our ancillary study, 185 agreed to MRS. We obtained usable hepatic fat data from 151 (82%) and ATP data from 105 (58%). Obesity was the strongest predictor of failed data acquisition; every unit increase in the body mass index reduced the likelihood of successful fat data by 11% and ATP data by 14%. The inter- and intrarater reliability were excellent for fat (intraclass correlation coefficient=0.99), but substantially more variable for ATP. Fat measures appeared relatively stable, but this was less true for ATP. CONCLUSIONS Obesity can hinder (1)H and (31)P MRS data acquisition and subsequent analysis. This impact was greater for hepatic ATP than hepatic fat.