The aim of this study was to statistically evaluate the effects of some formulation and process variables in the spheronization of microparticulates of ibuprofen using the rotor disk fluid-bed technology and water as binder. Preliminary studies revealed that presence of surfactant, plate material type, and nature and content of binder influenced the process and quality of the spheronized material. A 2 x 2 x 3 full factorial randomized experiment was designed, demonstrating the influence of these factors on properties such as percent yield, particle size distribution, densities, ibuprofen release, moisture content, etc., as well as their interactions in the experimental response. A response known as the usable fraction was created representing microparticulates of 250 to 850 microm sizes (mesh size 20-60). The reproducibility of the spheronization process was assessed by blocking the experiments with the experiments within the blocks randomly replicated. The main effects included two binder levels (X1), two surfactant levels (X2), and a three-level plate type (X3) in which 2 two-level factors were collapsed into a single three-level factor. The results from the statistical analysis (general linear model, JMP 4) showed that the variables studied had a significant influence on most of the response variables evaluated (p < 0.05), with the binder level proving to be the most significant of the three. There was also significant interaction (p<0.05) between binder level and plate type with the drug content, friability, sphericity, loss on drying (LOD), and usable fraction response variables, and between the binder and the surfactant levels with the drug content, Q20, true density, geometric mean diameter, LOD, and usable fraction responses. High levels of surfactant and binder increased the sphere size, while low levels decreased it. Significant (p < 0.05) interaction was also observed between the plate type and surfactant level with the drug content, geometric mean diameter, and more or less with Q20 and bulk density response variables. Blocking of the experiments had no significant effect on the process and product characteristics analyzed, indicating the reproducibility of the process.