Abstract Photometric observations of the spectroscopically confirmed z ≈ 9.1 galaxy MACS1149-JD1 have indicated the presence of a prominent Balmer break in its spectral energy distribution, which may be interpreted as due to very large fluctuations in its past star formation activity. In this paper, we investigate to what extent contemporary simulations of high-redshift galaxies produce star formation rate variations sufficiently large to reproduce the observed Balmer break of MACS1149-JD1. We find that several independent galaxy simulations are unable to account for Balmer breaks of the inferred size, suggesting that MACS1149-JD1 either must be a very rare type of object or that our simulations are missing some key ingredient. We present predictions of spectroscopic Balmer break strength distributions for z ≈ 7–9 galaxies that may be tested through observations with the upcoming James Webb Space Telescope and also discuss the impact that various assumptions on dust reddening, Lyman continuum leakage and deviations from a standard stellar initial mass function would have on the results.