The identification and utilization of new germplasm for broadening the maize genetic base of a breeding program is a continuous challenge for maize breeders worldwide. In the current study, 25 tropical maize (Zea mays L.) inbred lines were crossed with six testers from three different heterotic group (Suwan1, Reid and non-Reid). Among the 25 tropical lines, there were sister lines derived from the 3^rd, 4^th, and 5^th generations of inbreeding in three different populations. The heterosis and combining ability of the 25 tropical maize inbred lines were examined at three locations in Yunnan province of Southwestern  China  in 2013 according to a NC-IImating design. The key results obtained from this study were: 1) six tropical inbreds have been identified that can be utilized directly in maize breeding programs in the Southwestern region of China; 2) greater variation in combining ability was observed when the lines were evaluated at earlier generations (S1-S3) versus at higher levels of inbreeding with S4 being a possible critical stage, suggesting that at this level of inbreeding many of the alleles affecting combining ability might already be fixed; 3) an effective way to improve local elite maize lines is to use tropical germplasm from Suwan1 to cross with germplasm from Reid and non-Reid groups in the following mating patterns of Reid×(non-Reid/Suwan1), or non-Reid×(Reid/Suwan1) or (Reid/Suwan1)×(non-Reid/ Suwan1) since the crosses in these patterns did not change the original lines’ heterotic groups after crossed with germplasm from the Suwan1 group; and 4) the 25 tropical maize inbred lines have been successfully classified into three maize heterotic groups based on both combining ability effects and grain yield, and further verified "the 3 heterotic groups" and "triangle" heterosis pattern for use in maize breeding for the Southwestern region of China.