Time-restricted feeding (TRF) initiated early during the dark phase prevents the metabolic consequences of a high-fat diet in rodent models. However, the metabolic consequences of delaying the initiation of TRF, akin to breakfast skipping in humans, is unclear. We assigned 8-week-old male C57BL/6J mice (n = 192) to chow or high-fat diet ad libitum (AL) for 4 weeks, before randomization to continue AL or 10 h of TRF, initiated at lights off (TRFe) or 4-h after lights off (TRFd) for a further 8 weeks. Oral glucose tolerance tests (1 g/kg), metabolic monitoring and body composition by echoMRI were performed, and tissues were collected at six time points. TRF reduced weight and fat mass vs AL, with a greater reduction in TRFe vs TRFd. TRF improved glucose tolerance and protected mice from high-fat diet-induced hepatosteatosis vs AL, with no difference between TRFe and TRFd. TRF increased the amplitude of Bmal1, Cry1, Per2, Nampt, and Nocturnin mRNA levels in liver. A phase delay in Bmal1, Cry1, Per2, Reverbα, Nampt, NAD, Sirt1, and Nocturnin was observed in TRFd. Thus, delaying TRF limited the weight benefit and induced a phase delay in the hepatic clock, but improved metabolic health. Allowing more flexibility in when TRF is initiated may increase the translational potential of this dietary approach in humans.