In this study, a simplified semi-physical retrieval algorithm for latent heat (LH) released from precipitation over the Tibetan Plateau (TP) is derived and analyzed. The physical basis of this algorithm is that the vertical gradient of rain rate (-dR/dZ or Γ) represents the temporal rate of rain formation based on the steady-state assumption, and the precipitation formation rate is closely related to the cloud formation rate, which is directly proportional to the latent heating rate. In this algorithm, the LH rate is represented as a linear function of Γ with fixed slope and intercept term determined by 3-month Weather Research and Forecasting Model simulations over the TP. Comparison to model results shows that the retrieval scheme can correctly capture the main features of LH horizontally and vertically. Comparison with results from other two widely accepted LH algorithms using Global Precipitation Measurement Dual Precipitation Radar real observations shows that this retrieval scheme generally agrees with them over low-altitude areas but yields more convective-type LH over the highlands with a relatively lower heating center.