A novel algorithm for attenuation calibration in distributed temperature sensors (DTS) based on Raman scattering is presented and experimentally demonstrated, for the first time to the best of our knowledge. Numerical simulation is implemented to explore the signal distribution of Stokes and anti-Stokes lights along sensing fiber with the traditional DTS configuration. The proposed attenuation calibration method is to process the directly detected signal related to the temperature profile and calculate the calibration coefficient to maintain the linearity of temperature measurement and its uniformity along the whole fiber. The experimental results indicate that the attenuation issue of the signal along the sensing fiber is solved simply and efficiently. Simultaneously, a good linearity of the temperature measurement is achieved and the sensitivity remains uniform in the measuring range. The enhanced DTS system reduces the maximal error in temperature measurement from 18.01°C to 1.56°C along an approximate 10 km sensing fiber. The simplicity, validity, and reliability of the proposed method make the DTS system a better candidate in practical applications without the extra requirements of the existing system configuration.