Tropical reef corals are well known for their sensitivity to rising temperature, yet surprisingly little is known of the mechanisms through which temperature acts on intact coral colonies. One such mechanism recently has been suggested by the association between the growth of juvenile corals and seawater temperature in the Caribbean, which suggests that temperature causes a transition between isometric and allometric growth scaling in warmer versus cooler years, respectively (Edmunds in Proc R Soc B 273:2275-2281, 2006). Here, this correlative association is tested experimentally for a cause-and-effect relationship. During April and May 2006, juvenile colonies (8-35 mm diameter) of massive Porites spp. from Moorea, French Polynesia, were incubated at warm (27.8 degrees C) and cool (25.7 degrees C) temperatures for 15 days, and their response assessed through the scaling of growth (change in weight) with colony size. The results reveal that the scaling of colony-specific growth (mg colony(-1) day(-1)) was unaffected by temperature, although growth absolutely was greater at the cool compared to the warm temperature, regardless of colony size. This outcome was caused by contrasting scaling relationships for area-specific growth (mg cm(-2) day(-1)) that were negatively allometric under warm conditions, but independent of size under cool conditions. In April 2007, a 22 days field experiment confirmed that the scaling of area-specific growth in juvenile Porites spp. is negatively allometric at a warm temperature of 29.5 degrees C. Based on strong allometry for tissue thickness, biomass, and Symbiodinium density in freshly collected Porites spp., it is hypothesized that the temperature-dependency of growth scaling in these small corals is mediated by the interaction of temperature with biomass.