Different parvoviruses are used interchangeably as models in validation studies to demonstrate effective clearance of small viruses by filtration in the manufacturing of biotherapeutics. The aim of these experiments was to determine if filtration of different parvoviruses (canine parvovirus [CPV], minute virus of mice [MVM], and porcine parvovirus [PPV]) results in similar virus retention. While filtration with a Planova™ 20 N filter (mean pore size: 19 ± 2 nm) completely removed PPV and MVM from the filtrate (mean log reduction factors [LRFs] ≥5.8 to ≥7.3 log10), CPV was only partly retained (3.6 log10) in a series of single and co-spike experiments. Additional co-spike experiments in 2 different feedstreams using 10 commercially available small pore filters confirmed these results; the LRF for CPV was around 2 log10 lower than for MVM and PPV. A sizing study using filters with mean pore sizes between 16.5 and 19 nm resulted in complete removal of CPV only with smaller pore sizes (17 and 16.5 nm). CPV behaves differently to MVM and PPV in viral filtration due to its apparent smaller size, suggesting CPV represents a worst-case model for other parvoviruses. Interpretation of efficacy and robustness of virus filtration thus depends on the choice of model virus.