Corrosion is a main cause of failures in the gas pipeline. Both predicting pipeline failures and designing a maintenance plan play a key role in effective use of energy and security of civil life. This paper deals with model-based reliability estimation of corroded pipelines. Because a pipeline includes uncertainty in its operation, a statistical method called first-order and second- moment (FOSM) is adopted in this paper. One of distinctions in our work is to divide the whole pipeline into several clusters. Because magnitudes and likelihoods of metal corrosion defects depend on the pipeline commission environment, such a feature makes it easier to obtain practical reliability predictions. To this end, a fuzzy logic based clustering technique is employed. Numerical experiments are conducted based upon a modified field dataset. The result indicates that our proposed scheme provides more advisory predictions on the pipeline failure which is useful for preventive maintenance.