Thermal convective precipitation (TCP) is a usual weather phenomenon under subtropical high. TCP always occurs suddenly with small temporal and spacial scales, and it is a difficult point in the current weather forecasting operations. In this article, the convective temperature (Tc) is used to forecast TCP. First, the algorithm of Tc in MICAPS 3 is improved, and then the sounding and surface observation data from Nanjing Station in July and August during 2004-2013 are used to discuss the feasibility of forcasting TCP depending on Tc under the control of subtropical high. The results show that, the average probability of TCP occurrence at Nanjing Station is 1/6. The probability of TCP occurrence increases and then decreases with the change of temperature difference between the day high temperature Tmax and Tc. When Tmax-Tc is smaller than -0.5℃, the probability of occurrence is lower than average probability. When Tmax-Tc is 0.5 to 1.5℃, the probability of TCP occurrence is the highest, reaching 40%. The two requirements for TCP occurrence under the control of subtropical high are Tmax > 30℃ and Tmax-Tc > -3.5℃. In addition, statistics of 34 TCP cases show that TCP mainly occurs from 13:00-18:00 BT every day and its duration is 50 min averagely with the mean precipitation being 7.8 mm.