Thermal stress has been considered to be among the most important determinants of organismal distribution in the rocky intertidal zone. Yet our understanding of how body temperatures experienced under field conditions vary in space and time, and of how these temperatures translate into physiological performance, is still rudimentary. We continuously monitored temperatures at a site in central California for a period of two years, using loggers designed to mimic the thermal characteristics of mussels, Mytilus californianus. Model mussel temperatures were recorded on both a horizontal and a vertical, north-facing microsite, and in an adjacent tidepool. We periodically measured levels of heat shock proteins (Hsp70), a measure of thermal stress, from mussels at each microsite. Mussel temperatures were consistently higher on the horizontal surface than on the vertical surface, and differences in body temperature between these sites were reflected in the amount of Hsp70. Seasonal peaks in extreme high temperatures ("acute" high temperatures) did not always coincide with peaks in average daily maxima ("chronic" high temperatures), suggesting that the time history of body temperature may be an important factor in determining levels of thermal stress. Temporal patterns in body temperature during low tide were decoupled from patterns in water temperature, suggesting that water temperature is an ineffective metric of thermal stress for intertidal organisms. This study demonstrates that spatial and temporal variability in thermal stress can be highly complex, and "snapshot" sampling of temperature and biochemical indices may not always be a reliable method for defining thermal stress at a site.