Chemical sensors based on metal oxides (MOx) are most promising for emerging applications including medical breath analysis, distributed environmental monitoring and rapid food quality assessment. Yet, such sensors are not established in daily practice, mainly due to their limited selectivity, sensitivity and stability. Catalytic filters offer an effective solution to improve these by converting interferants to inactive species and/or target analytes to more responsive ones. This has been exploited successfully for alkane sensors, enabling their commercial utilization. Here, catalytic filters are discussed as promising tool to optimize the performance of chemoresistive MOx sensors. First, we provide an overview of chemical and physical parameters that govern the catalytic reactivity of such filters and we compare their implementation as overlayers and packed beds. Thereby, recent advances in the nanoscale design of suitable materials to finely tune their catalytic properties are elaborated. Next, filter solutions for analytes of various chemical families (including alkanes, alkenes, inorganics, alcohols, ketones and aromatics) are discussed and quantitatively compared also to other state-of-the-art detectors. Emphasis is placed on present challenging scenarios, for instance, the distinction of analytes from significantly higher concentrated interferants (e.g., breath markers in the presence of background ethanol in hospitals) or chemically similar compounds (e.g., benzene from xylene and toluene in air quality assessment). This is followed by examples demonstrating the integration of such filter-sensor concepts into devices and their evaluation under real conditions. Finally, opportunities and research frontiers are highlighted to inspire future research.