Necrotizing enterocolitis (NEC) is a life-threatening gastrointestinal disorder afflicting preterm infants, which is currently unpreventable. Fecal microbiota transplantation (FMT) is a promising preventive therapy, but the transfer of pathogenic microbes or toxic compounds raise concern. Removal of bacteria from donor feces by micropore filtering may reduce this risk of bacterial infection, while residual bacteriophages could maintain the NEC-preventive effects. We aimed to assess preclinical efficacy and safety of fecal filtrate transplantation (FFT). Using fecal material from healthy suckling piglets, we compared rectal FMT administration (FMT, n = 16) with cognate FFT by either rectal (FFTr, n = 14) or oro-gastric administration (FFTo, n = 13) and saline (CON, n = 16) in preterm, cesarean-delivered piglets as models for preterm infants. We assessed gut pathology and analyzed mucosal and luminal bacterial and viral composition using 16S rRNA gene amplicon and meta-virome sequencing. Finally, we used isolated ileal mucosa, coupled with RNA-Seq, to gauge the host response to the different treatments. Oro-gastric FFT completely prevented NEC, which was confirmed by microscopy, whereas FMT did not perform better than control. Oro-gastric FFT increased viral diversity and reduced Proteobacteria relative abundance in the ileal mucosa relative to control. An induction of mucosal immunity was observed in response to FMT but not FFT. As preterm infants are extremely vulnerable to infections, rational NEC-preventive strategies need incontestable safety profiles. We show in a clinically relevant animal model that FFT, as opposed to FMT, efficiently prevents NEC without any recognizable side effects.