The mummichog Fundulus heteroclitus is a trophically important fish inhabiting Atlantic coastal salt marshes, with few in situ estimates of overwinter survival throughout the species range. We estimated overwinter apparent survival rates of F. heteroclitus at the approximate mid-latitudinal species range [coastal North Carolina (USA)] in four tidal creeks that experience variable winter water temperatures. To estimate apparent survival, we fitted a Cormack-Jolly-Seber model to daily mark-resight data autonomously obtained from fish marked with passive integrated transponder tags. Creek, year, mean daily water temperature, change in mean daily temperature, fish length and fish condition were considered for effects on the modelled parameters: apparent survival (Φ) (product of true survival and site fidelity) and detection probability (p). Modelling showed that water temperature and fish metrics were not related to Φ. Water temperature was directly related to p, indicating reduced fish activity and thus reduced detection probability or poor antenna detection performance at low temperatures. Creek was related to Φ and p, and the creek most open to its downstream estuary (lacking a culvert) had lower rates than the others. Greater loss (fish mortality plus emigration) in this one creek may more effectively transfer production of F. heteroclitus to larger waterbodies via emigration or predation. Conversely, lower Φ may reflect reduced detection efficiency. The results suggest that F. heteroclitus survival is insensitive to variable winter water temperatures typical of thermal dynamics in shallow estuaries in this region of its range. Median creek-specific overwinter Φ rates (range of median values, 2 × 10⁻⁸, 0.04) were roughly equal to previously published rates for these creeks during the growing season (April–October). At these latitudes and with increasingly moderate winters, the results indicate that natural mortality could arise equally or more so from predation during the growing season than mechanisms such as starvation, direct mortality, thermal morbidity and stress-related susceptibility to predation resulting from intermittently low water temperatures during the overwinter season.