1.Behavioural adaptations of ectotherms to thermally heterogeneous environments are still overlooked in the literature despite the fact that organismal behaviour could enhance survival in the warming world. This is particularly critical in the intertidal where most ectotherms live at, or near to the upper limit of thermal tolerance. 2.This study investigated (i) the environmental factors determining the body temperatures of the intertidal gastropod Nerita atramentosa, (ii) the space-time variability in environmental and individual body temperatures and (iii) the potential variability in N. atramentosa thermoregulatory behaviours, i.e. microhabitat selection and aggregation. 3.Thermal imaging was used to assess the body temperatures of N. atramentosa and surrounding substrata over two seasons (autumn and summer), at two shore levels (low- vs. high-shore levels) within two habitats of different topographic complexity (rock platform and boulders) on the same rocky shore. 4.Snail body and substratum temperatures were significantly and positively correlated within each habitat at both seasons. Substratum temperature may thus be considered as a primary driver of body temperatures of organisms that attach to a substratum. Substratum temperature and other variables such as solar irradiance critically need to be integrated in climate-change models that use single climatic variables (e.g. air temperature) that are not necessarily correlated with individual body temperatures in nature. 5.The high space-time variability in both substratum and body temperatures reinforces the growing evidence that small spatial scale variations may surpass those observed at larger spatial scales. 6.Nerita atramentosa thermoregulatory behaviour under high thermal stress appeared to be habitat specific. 7.The small spatial scale heterogeneity in environmental and individual temperatures and in thermoregulatory behaviours has stressed the need to focus on body temperature patterns at the niche level and to integrate the organismal behaviour in climate-change models.