Accurate stem taper functions are necessary for estimating stem diameter, form and tree volume, and are important for wood estimation and timber utilisation. Betula alnoides is a valuable plantation hardwood species under large-scale development in South-East Asia and south China, but no study on its stem taper has been reported yet. Here, 28 well-known taper functions from three groups of models (single, segmented and variable-form taper functions) were fitted and validated separately with diameter-height datasets from 90 and 29 trees of B. alnoides. All trees were sampled from its 8-36-year-old even-aged plantations in south China. Non-linear regression techniques were applied to estimate the parameters in the taper equations using the fitting data set. Then bias, absolute bias, mean squared error and per cent explained variation were applied to examine model accuracy. Student's paired t-tests for observed and predicted diameters were used to examine the validity of these functions based on the validation data. Of the 28 functions, the Bruce, Max-Burkhart and Muhairwe models were the best single, segmented and variable-form taper equations respectively. In general, variable-form taper equations performed better than simple and segmented functions. The Muhairwe model best estimated stem taper without obvious multicollinearity for this species.