Annual density banding in reef corals was discovered by Knuston et al in 1972. X ray photographs of coral skeleton show its banding structure with more information of the past climates and ocean environments on the basis of annual variation, it is easy to date the position along with the growth axis of the corals. Annual density banding patterns can trace sea surface temperature （SST）. Annual density variations obviously record cycling of one or more environmental signals. SST is the dominant factor on coral growth at the regions, where annual change of temperature is large. Annual growth length of corals and calcification rate can be used as proxies of annual average SST. The reef coral samples （SY 7 core） were collected at Xisha Yongxing Island （16°50′N, 112°20′E）, South China Sea, where both of annual change and season change of SST are large due to the influence of the monsoon. It is available to use the annual density banding of coral for reconstructing the past SST. The experimental procedures show as follows: First, a 10 mm thick longitudinal slice was cut along the central of section of the coral core. Secondly, the X ray photographs of the slice were taken by a medical X ray set. At last, the values of image density were gained by using a microdesitometer along with the growing axis at the step of 0 25 mm and scan area of 0 25 mm×0 25 mm. The length of scan step and the number of scan dots can be used as the proxies of annual growth rate and annual calcification rate respectively. Growth rate can be expressed by the Equation: L=0 25×n/β . L is annual growth rate （mm/a）; n is the number of scan dots per year; β is image scale; 0 25 is the length of scan step. Calcification rate proxy can be expressed by the Equation: G′=? 苮ni=1V i/β . G′ is annual calcification rate proxy; V i is the value of image density at i dot; β is image scale. Data of image density of X ray photograph, measured by use of microdesitometer in the period of 1937¹993, were derived for annual growth rate and annual calcification rate proxy. Both parameters were positively related to annual mean sea surface temperature, which was measured by the instruments at the oceanographic station. Growth rate thermometer and calcification rate proxy thermometer were set up using linear regression method. Both thermometers show: SST=0 1175× L+26 r=0 91 . Where SST is annual mean SST （℃）; L is annual growth rate （mm/a）; r is correlative coefficient; SST=0 0204× G′+26 r=0 85 . Where SST is annual mean SST in year （℃）; G′ is annual calcification rate proxy; r is correlative coefficient. We can use the growth rate thermometer and the calcification rate proxy thermometer to reconstruct the past sea surface temperature records. The annual mean error between two SST records calculated by the two thermometers is ranged in (0 12℃. Annual SST records in South China Sea showed that SST was warmer in 1940s, 1980s to 1990s than that in 1950s to 1970s. The abnormal SST records can clearly reflect the influence of the ENSO phenomenon too.