A novel approach using a 2-D population balance model is developed and applied to the analysis of experimental tracer crystal data. This approach is effective in discriminating among various functional forms of agglomeration kernel and enables estimation of the agglomeration kinetics. At present, the analysis is restricted to three simple agglomeration kernels and shows that the size-independent kernel best describes the agglomeration of Al(OH)(3) crystals during precipitation in caustic aluminate solutions. This agrees with the findings of Ilievski and White (1994). Estimates of the agglomeration kinetic parameters from the tracer data agree well with the experimentally observed values.