Abstract: In forest management, to predict the stem diameters inside bark are more important than the diameters outside bark. Measurement on diameter inside bark is both expensive and time-consuming, moreover, causing larger measurement errors especially for measuring standing trees. In this study, three types of model are compared for predicting inside bark diameters using stem analysis data of dahurian larch (Larix gmelinii Rupr.): Grosenbaugh's ratio equations, regression models, and taper function. Grosenbaugh's ratio equations had great flexibility, need neither parameters nor model fitting. The results of overall evaluation and comparisons of different sections indicated that the regression models had smaller prediction error, especially the model including diameter outside bark, total height, relative height, breast height diameters outside bark and inside bark by Cao and Pepper. The taper function had larger prediction error of diameters inside bark but did not require outside bark diameters variables. All these models have some adaptability in the forest management process.