Ostracod assemblages and the δ18O and δ13C records of Cytheridella ilosvayi provide a ∼12,000 yr paleohydrological reconstruction of Little Salt Spring, a sinkhole lake and underwater archaeological preserve in west central Florida. The ostracod record documents relative changes in water temperature and hydrologic characteristics of the lacustrine system that are the result of warming air temperatures, changes in the relative contributions of input waters (shallow vs. deep groundwater), rainfall patterns and relative sea level rise during the Holocene. The data indicate that LSS was initially supplied by relatively low amounts of direct rainfall and surface runoff as the spring's location was closer to the recharge area and far from the freshwater/saltwater interface. At about 11,000 yr BP, LSS became groundwater supported, initially by a shallow freshwater aquifer with low δ18O composition until ∼5700 yr BP, and after that time by 18O-enriched and increasingly mineralized groundwater originating from a deeper, carbonate aquifer as the regional water table rose bringing the saltwater interface closer to the location of LSS which at this time was becoming a groundwater discharge area. The data also shows that the most abrupt and pronounced hydrologic changes occurred during the Late Holocene with an interval of low δ18O values (∼− 2.86‰) between 2700 and 2000 yr BP, followed by a stepped increase in δ18O composition with maximum values (∼− 1.46‰) between 1100 and 900 yr BP. Over these periods to the present time the ostracod assemblage is characterized by brackish species (Cyprideis spp. and Limnocythere floridensis). It is inferred that the interval of decreased isotopic values indicates mixing with saline water during dry climatic conditions, while the interval of more enriched isotopic values suggest saltwater intrusion during a period of, presumably, a sea level highstand. Although some discrepancies exist between other records from the northern Caribbean region and the one from LSS with regards to the exact timing for the Late Holocene drying and subsequent increase in rainfall conditions, the differences between the records may be explained by greater continental influences on ocean–atmosphere interactions on the Florida peninsula, a highly complex groundwater mixing balance in LSS, and by a better constrained age-model in the other records.