Pattern-based fuzzy predictive (PFP) control, proposed for processes with large time-delay, avoids traditional process dynamic models while providing effective disturbance rejection. However, chemical processes often undergo considerable uncertainties, which will affect the performance of the PFP scheme. An extended PFP (EPFP) control scheme is proposed to account for process dynamics with uncertainties. It incorporates a first-order lag unit, viewed as a special filter, with the original PFP, allowing the designer to set a trade-off between performance and robustness. In practice, the filter parameter tuning can be carried out automatically through on-line parameter optimization. This approach is applied to time-delay compensation for a time-delayed chemical process with uncertainties. Robustness of the EPFP is analyzed via Bode plots under process uncertainties. Illustrative examples are provided to show the effectiveness of the scheme to control chemical processes with time-delay and uncertainties.