This study aimed to determine the spatial autocorrelation patterns of tree growth and the mechanisms influencing habitat heterogeneity. A dataset of observed forest growth was used, with data between 2009 and 2014 from a 21.12 hm² stem-mapped mixed forest plot (temperate coniferous and broadleaved forest) in Jiaohe, Jilin Province. Based on habitat type classifications, Ripley's L(r) function was adopted to evaluate the spatial distribution characteristics of tree locations. A mark correlation function (MCF) was used to evaluate the spatial autocorrelation characteristics of tree growth. Our results showed that (1) Pinus koraiensis (habitat type 3: 1-5 m), Quercus mongolica (habitat type 3: 1-3 m), Juglans mandshurica (habitat type 2: 1-2 m; habitat type 3: 1-7 m), Phellodendron amurense (habitat type 2: 1-3 m; habitat type 4: 1-5 m), Fraxinus mandshurica (habitat type 3: 1-2 m; habitat type 4: 1-2 m), and Euonymus verrucosus (habitat type 2: 1-15 m) were randomly distributed at particular scales and for certain habitat types. Nonetheless, they mainly showed clumped distribution patterns. For ten other species, the spatial distribution of tree positions showed an aggregated pattern at a scale of 0-30 m. (2) The mark correlation function analysis indicated that the radial growth of Ulmus davidiana var. japonica, Corylus mandshurica, Acer mono, E. verrucosus, and Carpinus cordata were positively autocorrelated for at least one of the habitat types. In contrast, the radial growth of Syringa reticulata var. amurensis, J. mandshurica, Ulmus laciniata, E. verrucosus, F. mandshurica, Tilia amurensis, Tilia mandshurica, C. mandshurica, A. mono, and Acer mandshuricum were negatively autocorrelated at least in one of habitat types. Other species did not show any significant autocorrelations at any of the studied scales. Our results indicate that different species had different spatial autocorrelation characteristics of radial growth, as well as significantly different habitat dependencies.