There are presently two main avenues in the stochastic modeling of depositional facies: pixel-based and object-based geostatistics. They both have strengths and weaknesses: traditional pixel-based geostatistics is good at data conditioning, but it depends on variograms to capture spatial structures and hence fails to reproduce definite patterns common to most geological facies; while object-based geostatistics is good at reproducing crisp facies shapes but is difficult to condition to dense well data or exhaustive 3D seismic data. Multiple-point simulation, a newly developed pixel-based technique, integrates the strengths of both: it keeps the flexibility of pixel-based techniques for data conditioning, while allowing pattern reproduction through consideration of multiple-point statistics. In this paper, a workflow for multiple-point stochastic simulation is discussed in details. This workflow is applied to an industry project. The results show reproduction of the prior geological knowledge and honoring of both well and seismic data.