Abstract: The objectives of this study were to develop the equation systems of branch diameter, branch length and branch angle for natural Dahurian larch (Larix gmelinii) using the error-in-variable simultaneous equation approach, and to compare four parameter estimation approaches. The data were collected from 2306 branch samples of 36 trees in Songling Forest Bureau in Daxing'anling.The results showed that the Seemingly Unrelated Regression (SUR) approach could explain the correlation of error terms across the equations and decrease standard error of parameter estimation. Equation systems of branch characteristics established in our study show better precision and can better describe variation of branch characteristics and crown profile. In addition, our study also developed equation systems of branch diameter, branch length and branch angle with stand density variables using nonlinear mixed-effects modeling techniques. Stand density has a significant impact on branch diameter, branch length, branch angle and crown profile based on significant parameter estimates and better precision.