Developing Salix matsudana flowering phenological model based on temperature and photoperiod is crucial for finding the main meteorological factors which affected the spatio-temporal changes of the Salix matsudana flowering dates, revealing the ecological mechanism of regulating plant flowering dates, and providing reference information for improving human health problems such as environmental pollution and pollen allergy caused by willow catkins. In this study, we used Salix matsudana phenology data of beginning of flowering (BF), full flowering (FF) and end of flowering (EF) at 49 phenological observation stations acquired from the China Meteorological Administration (CMA) phenological network during 1982-2011 to establish and compare six flowering models (simple accumulated temperature model, three-base temperature model, eight-stage temperature model, simple accumulated temperature×photoperiod model, three-base temperature×photoperiod model, eight-stage temperature×photoperiod model). According to external validation results, we selected the optimal phenological model for BF, FF and EF, respectively. Subsequently, we obtained spatial patterns of multi-year and yearly Salix matsudana flowering dates over continuous geographic coverage during 1982-2011 by using the optimal phenological model. The results showed that Salix matsudana flowering was better fitted to the model which considered both the effect of temperature and photoperiod. The eight-stage temperature×photoperiod model outperformed the other two models for BF and FF. For EF, the three-base temperature×photoperiod model performed better than the other five models. It indicated that photoperiod and temperature were the main meteorological factors influencing the Salix matsudana flowering. The effect of simulation and prediction was more accurately for different stations and years by using the optimal flowering phenological model. The multi-year simulated mean dates of BF, FF and EF were April 24, April 28, and May 3, respectively. And the multi-year simulated mean Salix matsudana flowering dates showed a delayed spatial progression from low altitude to high altitude, from south to north, and from west to east. The simulated linear trends in BF, FF and EF dates in most areas from 1982 to 2011 indicated an advancement and the areas with significant advancement accounted for 49.78%, 50.01% and 53.40% of the total area, respectively. In addition, the multi-year simulated mean LF was 9 days, and no significant trends were found in most areas for FL.