The predictability of precipitation is limited due to the important role finer scale processes play. However, demand for medium- (3–10 days) and extended-range (10–30 days) precipitation forecasts by agriculture, forestry, livestock, and water resource management users has grown significantly. In this study, a statistical post-processing technique combining Analog Post-processing (AP) and Probability-Matched mean (PM), called APPM, is developed to perform bias correction and downscaling for 1–14 day precipitation forecasts in Taiwan. The aim is to provide users with more accurate Quantitative Precipitation Forecasts (QPF) and more reliable Probabilistic Quantitative Precipitation Forecasts (PQPF). Analog Post-processing (AP) searches for the best analogs to the current forecast in a historical set of predictions. The AP forecast ensembles are derived from the observed high-resolution precipitation patterns corresponding to the historical forecast analogs that most resemble the current ensemble precipitation forecast. Frequency counting is then applied to the AP ensembles to produce calibrated and downscaled 1–14 day PQPF. For QPF with a more realistic range of precipitation amounts, PM is applied on the AP ensemble mean. Forecast evaluation shows that the raw ensemble is under-dispersive with a wet bias. In contrast, the AP ensemble spread well represents the forecast uncertainty. Compared to the raw PQPF, the AP-based probabilistic forecasts have better reliability, higher skill in discrimination, and higher economic value for a wider range of users. The calibrated QPF displays finer scale details of precipitation, explains about 3 to 5 times more variance in observations, as well as removes most bias and reduces Mean Absolute Error (MAE) in most seasons and lead times compared to the raw QPF.