Studies into the viral fraction of complex microbial communities, like in the mammalian gut, have recently garnered much interest. Yet there is still no standardized protocol for extracting viruses from such samples, and the protocols that exist employ procedures that skew the viral community of the sample one way or another. The first step of the extraction pipeline often consists of the basic filtering of macromolecules and bacteria, yet even this affects the viruses in a strain-specific manner. In this study, we investigate a protocol for viral extraction based on ultrafiltration and how the choice of ultrafilter might influence the extracted viral community. Clinical samples (feces, vaginal swabs, and tracheal suction samples) were spiked with a mock community of known phages (T4, c2, Φ6, Φ29, Φx174, and Φ2972), filtered, and quantified using spot and plaque assays to estimate the loss in recovery. The enveloped Φ6 phage is especially severely affected by the choice of filter, but also tailed phages such as T4 and c2 have a reduced infectivity after ultrafiltration. We conclude that the pore size of ultrafilters may affect the recovery of phages in a strain- and sample-dependent manner, suggesting the need for greater thought when selecting filters for virus extraction.