Massive corals provide a useful archive of environmental variability, but careful testing of geochemical proxies in corals is necessary to validate the relationship between each proxy and environmental parameter throughout the full range of conditions experienced by the recording organisms. Here we use samples from a coral-growth study to test the hypothesis that Sr/Ca in the coral Siderastrea siderea accurately records sea-surface temperature (SST) in the subtropics (Florida, USA) along 350 km of reef tract. We test calcification rate, measured via buoyant weight, and linear extension (LE) rate, estimated with Alizarin Red-S staining, as predictors of variance in the Sr/Ca records of 39 individual S. siderea corals grown at four outer-reef locations next to in-situ temperature loggers during two, year-long periods. We found that corals with calcification rates < 1.7 mg cm⁻² d⁻¹ or < 1.7 mm yr⁻¹ LE returned spuriously high Sr/Ca values, leading to a cold-bias in Sr/Ca-based SST estimates. The threshold-type response curves suggest that extension rate can be used as a quality-control indicator during sample and drill-path selection when using long cores for SST paleoreconstruction. For our corals that passed this quality control step, the Sr/Ca-SST proxy performed well in estimating mean annual temperature across three sites spanning 350 km of the Florida reef tract. However, there was some evidence that extreme temperature stress in 2010 (cold snap) and 2011 (SST above coral-bleaching threshold) may have caused the corals not to record the temperature extremes. Known stress events could be avoided during modern calibrations of paleoproxies.