This article deals with the optimal short-term scheduling of multistage batch plants with parallel units and sequence-dependent changeovers, together with the optimal selection of the number and size of batches to be produced. A new resource-task network-based, multiple time-grid continuous-time formulation is proposed that explicitly considers a virtual, shared, intermediate storage unit per stage to keep track of material transfer between processing units belonging to consecutive stages of production. Adequate material transfer is implicitly ensured through mass balances and timing constraints relating the times of event points of dissimilar grids. The new formulation is compared with a conceptually different approach from another research group. The results for several example problems show that the new formulation is tighter, typically requiring fewer event points to find the global optimal solution, and is significantly more efficient computationally. The results also show that for single batch problems other approaches are preferable. (C) 2009 American Institute of Chemical Engineers AIChE J, 55: 2122-2137, 2009