In tunnel engineering, the selection of tunnel support patterns should be estimated accurately to ensure stability of the tunnel, which may be caused by unexpected hazardous zones ahead of tunnel face. This study presents a method to estimate the selection of support patterns using artificial neural network (ANN) based on 318, 649 Measurement While Drilling (MWD) data. Controlled trials are conducted considering different input layer sizes and hidden layer sizes to obtain the optimal ANN model. Combinations of 6 feature parameters including penetration rate (PR), hammer pressure (HP), rotation pressure (RP), feed pressure (FP), hammer frequency (HF) and specific energy (SE) correspond to the different input layer sizes of the ANN. Average accuracy (A), average computing-time (T), sensitivity and stability are adopted as the performance index. The results show that a strong correlation exists between MWD data and support patterns. The combination of 6 feature parameters outperforms the subset of the entire feature parameters in terms of A, sensitivity and stability. The ANN model with the combination of PR, HP, RP, FP, HF and SE as the input feature parameters has the highest estimation stability. The ANN model with 6 feature parameters and one hidden layer with 30 nodes is proposed as optimal model considering all indices. The results confirm that it is feasible to estimate support patterns ahead of tunnel face using ANN based on MWD data.