In　this　paper,　error　function　was　proposed　for　the　first　time　to　predict　the　normalized　flux　in　dead-end　microfiltration　process　at　different　operating　conditions　(transmembrane　pressure,　concentration,　and　temperature)　using　various　feed　suspensions　and　membranes.　The　results　showed　that　the　predictions　of　proposed　model　had　good　agreements　with　experimental　data.　In　addition,　model　parameter　k　is　analyzed　quantificational　by　using　multivariate　nonlinear　regression　method.　Moreover,　the　proposed　model　also　showed　high　predictive　accuracy　(2.88%)　than　cake　models　(4.48%),　and　two　empirical　models　(A　(6.19%)　and　B　(35.7%)).