Wood is a renewable material ideal for environmentally friendly buildings, but wooden building envelopes may face mold growth risks across different climates. To ensure the long-term service life of wooden buildings in China, it is imperative to evaluate the mold growth risk in each region. Nevertheless, large-scale regional assessments require significant effort and time. This study proposes a method based on meteorological data and a neural network (NN) model for regional mold risk assessment in light wood-framed wall envelopes. The NN model, comprising a one-dimensional convolutional neural network (1D-CNN) and long short-term memory (LSTM), is trained on meteorological data from the hot summer and cold winter (HSCW) region, which is one of China's five climatic regions. It is validated using simulated data and one year of field monitoring data. Finally, the model predicts time series of relative humidity and temperature with a mold index from the empirical VTT model to assess mold growth risk in the HSCW region. The validation results with simulated data show good performance, with average R2 values of 0.969 and 0.984 for predicting interior wall relative humidity and temperature, respectively. However, validation with monitoring data shows a decline in performance due to real-world complexities. The results of the risk assessment indicate that the wall used in this study is commonly at risk in the HSCW region. The proposed method is suitable for assessing mold risk in walls across diverse regional climates, thereby providing tailored improvements to the hygrothermal performance of walls.