The use of diving animals as autonomous vectors of oceanographic instruments is rapidly increasing, because this approach yields cost-efficient new information and can be used in previously poorly sampled areas. However, methods for analyzing the collected data are still under development. In particular, difficulties may arise from the heterogeneous data distribution linked to animals' behavior. Here we show how raw temperature data collected by penguin-borne loggers were transformed to a regular gridded dataset that provided new information on the local circulation off Kerguelen. A total of 16 king penguins (Aptenodytes patagonicus) were equippedwith satellite-positioning transmitters and with temperature–time–depth recorders (TTDRs) to record dive depth and sea temperature. The penguins' foraging trips recorded during five summers ranged from 140 to 600km from the colony and 11,000 dives 4100m were recorded. Temperature measurements recorded during diving were used to produce detailed 3D temperature fields of the area (0–200 m). The data treatment included dive location, determination of the vertical profile for each dive, averaging and gridding of those profiles onto 0.110.11 cells, andoptimal interpolation in both the horizontal and vertical using an objective analysis. Horizontal fields of temperature at the surface and 100m are presented, as well as a vertical section along the main foraging direction of the penguins. Compared to conventional temperature databases (Levitus WorldOcean Atlas andhistorical stations available in the area), the 3D temperature fields collectedfrom penguins are extremely finely resolved, by one order finer.