Projected temperature increases due to global climate change are likely to have localized impacts on the San Francisco Estuary (SFE). Increased water temperature in the SFE will lead to challenges for managing water resources. Many native species, such as salmon and smelt, rely on cooler water, and will be further stressed by increased water temperature, which may cause them to seek microrefugia. While several state and federal agencies in the SFE collect real-time water temperature data, most of the water temperature collection sites are at a fixed location or floating near the surface of the water column. This dataset includes four real-time water quality stations that provide water temperature data for both the surface and bottom positions in the water column. We obtained surface water temperature data from an integrated hourly water temperature dataset (https://doi.org/10.6073/pasta/7385985f68b02c0deb2a9e425a9f3ad8). This dataset included data downloaded from the California Data Exchange Center (CDEC; https://cdec.water.ca.gov/) and cleaned with a series of quality control (QC) checks (see integrated dataset metadata). We obtained bottom temperature data from the California Department of Water Resources (DWR) internal database Water Quality Portal (WQP). Data were integrated and standardized to hourly water temperature data in degrees Celsius, and the same series of quality control (QC) checks from the surface dataset were applied in a consistent manner to all stations. Bottom temperatures were selected from surface temperatures to provide measures of temperature difference. Datasets included in this package include source hourly surface and bottom data, both obtained from DWR’s WQP, as well as an integrated dataset of cleaned hourly surface and bottom data, with calculated surface-bottom temperature differences. Both datasets are filtered to the timeframe used in an analysis of surface-bottom temperature differences. Additionally, information regarding current and historical sensors used for water temperature data collection was obtained from station managers for each station and compiled in a metadata table.