The objectives of this study was to select nonlinear regression model of relationship between body parameters and live weight of swamp buffaloes for live weight estimation. The animals of Department of Animal Science, Faculty of Agriculture at Kampaeng Saen, Kasetsart University, Kampaeng Saen Campus were used in this study. A total of 172 animals were 41 male and 131 female 0 to 25 months old of swamp buffaloes. Parameters of body measurements were shoulder height (SH), hip height (HH), shoulder width (SW), hip width (HW), body length (BL) and hearth girth (HG) were measured in centimeters during weighting. Three of nonlinear regression models including exponential, polynomial quadratic and power models were analyzed relationship between body parameters and live weight in each body parameter. The results revealed that power model gave the best fit model of HG and live weight relationship with highest R sup(2) (0.9662, 0.9748 and 0.9702) in male, female and both sex of swamp buffaloes, respectively. In addition, percentage error and accuracy of 3 models were investigated between HG and live weight of swamp buffaloes (n=492). The result revealed that polynomial quadratic model showed highly accuracy (98.92 percent) between actual live weight and calculated weight from the equation of y = 0.0233x**(2) -2.9263x+129.81 with R sup(2) value of 0.9678, when y = estimated live weight (kg), x = heart girth (cm). In conclusion, this model provides a highly reliable and accurate method for estimating weights of swamp buffaloes using a single heart girth measurement which can be easily obtained with a girth tape in the field work.