In this paper, we present an observer design for online estimation of species concentrations in recirculation-based solid oxide fuel cell (SOFC) systems with integrated reformers. For the system considered, on-board reforming of methane results in mixture of several species of different concentrations along the fuel path. The presence of a fuel reformer gives way to coupling of system equations, in turn, increasing species interactions and complexity of the state equations. The knowledge of concentration of species at key locations in the fuel cell can help prevent cell damage and improve longevity. In this regard, the use of sensors to determine species concentration is an invasive process which is expensive to both utilize and maintain. While existing observers are designed either for chemical reactors or for a fuel cell exclusively, the proposed strategy aims to improve on that by considering a combined reformer and fuel cell and designing a nonlinear adaptive observer using readily measured concentrations and selected variables. For estimating certain critical indicators, such as fuel utilization, state transformations have been used in the design to obtain a more versatile and computationally efficient reduced order observer. The study develops detailed stability analysis of the observers and quantifies the effect of uncertainties on observer performance.