With the increasing interactions between natural gas systems (NGS) and power systems, component failures in one system may propagate to the other one, threatening reliable operation of the whole system. Due to neglect of such cross-sectorial failure propagation in integrated electricity-gas systems (IEGSs), traditional economy-oriented reserve expansion models may lead to unreasonable planning results. In order to address this, an innovative reserve expansion model is proposed to determine the allocation of energy production components through the harmonization between costs and reliability. First, novel multifactor-influenced reliability indices are defined considering synthetic effects of multiple uncertainties, including failure propagation, load uncertainties and generation failures. In reliability index formulation, contribution of failure propagation on system reliability is analytically expressed. To avoid high computational complexity, the fuzzy set theory is combined with conventional methods, e.g., Monte-Carlo simulation technique to reduce numerous contingency states. Sampled contingency states are aggregated into several clusters represented by a fuzzy number. To effectively solve the planning model, a decomposition approach is introduced and applied to decompose the original problem into a master problem and two correlated reliability sub-problems. Numerical studies show the proposed model can plan reasonable reserves to guarantee reliability levels of IEGSs considering failure propagation.