The karst forest in China is mainly located in the southwestern mountainous regions within the largest continuous karst landscape in the world. The representative subtropical mixed evergreen-deciduous broad-leaved forest in Maolan National Natural Reserve has a specific habitat, a complex community structure, and rich biodiversity. A major objective of ecological research is to infer the underlying ecological processes or mechanisms by analyzing spatial distributions of species. However, knowledge about species spatial distributions and underlying mechanisms in the karst forest is still limited. The concept of "fractal" has been used to describe the structure of vegetation at a wide range of spatial scales. Fractal dimension, as a parameter for quantication of the spatial structure of objects, is able to quantify the structure of complex objects (e.g. tree spatial distribution) in one single value, in terms of the object's spatial density and distribution. Using the box-counting dimension D_b and information dimension D_i of fractal dimension, as well as four kinds of pattern indexes (patchiness index, Green's index, dispersion index, and negative binomial parameter), we studied the spatial distribution patterns and the fractal properties of 30 tree species in the subtropical mixed evergreen-deciduous broad-leaved karst forest in Maolan National Natural Reserve, southwestern China. In addition, changes of the fractal dimension of the population's spatial distribution among different tree growth stages (sapling, pole, and adult) were analyzed. Correlations between two kinds of fractal dimensions, five kinds of population structure parameters, and four kinds of pattern indexes were also compared. Results showed that: 1) Among the 30 tree species, 28 species were spatially aggregated in their distribution except for Pittosporum crispulum and Diospyros oleifera. This prevailing aggregated pattern is closely related to the great heterogeneity of karst habitat and seed dispersal limitation; 2) D_b values of tree spatial distribution were between 0.589 and 1.870 while D_i values were between 0.498 and 1.711. Within a population, a high fractal dimension value indicates a strong ability to occupy ecological space and thus high aggregation. Therefore, the species can grow in the dominant position of a plant community. However, a low value reveals a weak occupation ability, low aggregation level, and associated position; 3) Averaged values of D_b and D_i in the spatial distribution of the 30 tree species from saplings to adult trees tended to decrease very significantly. This might be a reflection of the density-dependent effect resulting from the obvious decline of aggregation level and occupation ability with increased tree age; and 4) Correlation analysis showed a positive correlation between population abundances, importance values, and D_b and D_i values for these 30 tree species, indicating close relationships between changes of population abundance and importance value, occupation ability, and aggregation level. All of the kinds of pattern indexes, except for index of patchiness, had significant or very significant correlations to D_b and D_i, indicating that the aggregation level of population spatial distribution can be characterized by the fractal dimensions D_b and D_i. This study accurately quantified the spatial distribution patterns of trees in a karst forest by applying pattern indexes and fractal dimensions. Findings from this study contribute to a clearer understanding of the processes influencing the karst forest structure.