Skeletal cadmium-to-calcium (Cd/Ca) ratios in hermatypic stony corals have been used to reconstruct changes in upwelling over time, yet there has not been a systematic evaluation of this tracer's natural variability within and among coral species, between depths and across environmental conditions. Here, coral skeletal Cd/Ca ratios were measured in multiple colonies of Pavona clavus, Pavona gigantea and Porites lobata reared at two depths (1 and 7 m) during both upwelling and nonupwelling intervals in the Gulf of Panama (Pacific). Overall, skeletal Cd/Ca ratios were significantly higher during upwelling than during nonupwelling, in shallow than in deep corals, and in both species of Pavona than in P. lobata. P. lobata skeletal Cd/Ca ratios were uniformly low compared to those in the other species, with no significant differences between upwelling and nonupwelling values. Among colonies of the same species, skeletal Cd/Ca ratios were always higher in all shallow P. gigantea colonies during upwelling compared to nonupwelling, though the magnitude of the increase varied among colonies. For P. lobata, P. clavus and deep P. gigantea, changes in skeletal Cd/Ca ratios were not consistent among all colonies, with some colonies having lower ratios during upwelling than during nonupwelling. No statistically significant relationships were found between skeletal Cd/Ca ratios and maximum linear skeletal extension, δ13C or δ18O, suggesting that at seasonal resolution the Cd/Ca signal was decoupled from growth rate, coral metabolism, and ocean temperature and salinity, respectively. These results led to the following conclusions, (1) coral skeletal Cd/Ca ratios are independent of skeletal extension, coral metabolism and ambient temperature/salinity, (2) shallow P. gigantea is the most reliable species for paleoupwelling reconstruction and (3) the average Cd/Ca record of several colonies, rather than of a single coral, is needed to reliably reconstruct paleoupwelling events.