A Hybrid Artificial Neural Network (HANN) model for estimating the activity of a commercial Cu/ZnO/Al2O3-ZrO2 catalyst in a laboratory scale methanol steam reforming reactor has been presented. This model is also capable of predicting methanol conversion, selectivity and rate of hydrogen production. In the proposed model, the decay function of heterogeneous catalysts is combined with a feed-forward artificial neural network. To identify the activity of catalyst, a set of 96 data points during 1900 min of operation are obtained from the laboratory scale reactor. From these data, 56 points are selected for training (60%), 20 data points for testing (20%) and the remained ones for validating the developed hybrid network (20%). Results show that the HANN can appreciably predict the activity of the catalyst, and it is also capable of predicting conversion, selectivity and hydrogen production with the AAD% (average absolute deviation) of 1.296, 0.451 and 0.5816%, respectively. Finally by applying the proposed HANN model, process variables i.e. temperature and water to feed ratio are optimized such that by decreasing the activity of the catalyst, the conversion, selectivity and hydrogen production rate can be preserved as equal as the start of run (SOR) values.