To quantitatively measure tau aggregation in situ, we established a cell model system using a split green fluorescence protein (GFP) complementation assay. In this assay the more aggregated the protein of interest the lower the GFP fluorescence. Tau microtubule-binding domain constructs, whose aggregation characteristics have been described previously (Khlistunova et al. 2006), were used to validate the assay. The aggregation-prone construct exhibited the lowest GFP intensity whereas the aggregation-resistant construct showed the highest GFP intensity. To examine the role of glycogen synthase kinase 3 beta (GSK3 beta) activity and caspase 3 cleavage on tau aggregation, GFP complementation of full length (T4), caspase-cleaved (T4C3), and pseudophosphorylated at S396/S404 (T4-2EC) tau was examined in the presence of an active or a kinase-dead GSK3 beta. Extensive phosphorylation of T4 by GSK3 beta resulted in increased GFP intensity. T4C3 showed neither efficient phosphorylation nor a significant GFP intensity change by GSK3 beta. The GFP intensity of T4-2EC was significantly reduced by GSK3 beta accompanying its presence in the sarkosyl-insoluble fraction, thus demonstrating that T4-2EC was partitioning into aggregates. This indicates that if the majority of tau is phosphorylated at S396/S404, in combination with increased GSK3 beta activity, tau aggregation is favored. These data demonstrate that split GFP complementation may be a valuable approach to determine the aggregation process in living cells.