Fractal dimension (Df) has been extensively used for many years to characterize the morphological properties of aggregate systems. There are two main methods to estimate the fractal dimension of aggregates, namely the box-counting (BC) and power law (PL) methods. However, the relationship between the BC fractal dimension (Df, BC) and PL fractal dimension (Df, PL) has not been discussed yet. In this work, a series of three-dimensional aggregates with different input parameters (Df, PL and the number of primary particles) is generated by a tunable aggregation model. Then, the fractal dimensions (Df, BC, 3D) of all the aggregates are estimated by the 3D BC method. The relationship between Df, BC, 3D and Df, PL is investigated. We found that Df, BC, 3D is greater than Df, PL when Df, PL≤ 2.5. However, the situation is reversed when Df, PL> 2.5. Further, a novel projection method is proposed and applied to all the 3D aggregates to obtain their 2D projection images. In this projection method, the minimum projection overlapping area of all the primary particles in the aggregate is considered. Then, the fractal dimensions (Df, BC, 2D) of 2D projection images are estimated using the 2D BC method. Finally, correlations between Df, BC, 3D and Df, PL with Df, BC, 2D are established. Display omitted •Relationship established between BC and PL fractal dimension in 3D.•Novel projection method proposed to minimize the projection overlapping area.•Correlation between 2D BC and 3D PL fractal dimension established.