Aims Silage is grass, preserved by fermentation and used as winter feed for cattle. The impact of a range of current grass silage preparation practices on the survival of Escherichia coli C600φ3538(Δvtx2 ::cat) and on the induction, release and infectivity of free phage were investigated. Methods and results Wilted and fresh grass samples, from plots with and without slurry application, were ensiled with or without formic acid. Each treatment combination was inoculated with approximately 6 log10 CFU per g E. coli C600φ3538(Δvtx2 ::cat) (donor strain) and E. coli C600::kanamycinR (recipient strain) in test-tube model silos and incubated in the dark at 15°C. The physico-chemical (pH, ammonia, ethanol, lactic acid and volatile fatty acids) and microbiological (total viable counts, TVC, total Enterobacteriaceae counts, TEC, E. coli counts, ECC and lactic acid bacteria, LAB) properties of each fermentation were monitored throughout the experiment as were the concentrations of E. coli C600φ3538(Δvtx2 ::cat), E. coli C600::kanamycinR , free phage and transductants, using culture and PCR-based methods. Over the course of the experiment the pH of the grass samples typically decreased by 2 pH units. TVC, TEC and ECC decreased by up to 2·3, 6·4 and 6·2 log10 CFU per g, respectively, while the LAB counts remained relatively stable at 5·2-7·1 log10 CFU per g. Both donor and recipient strains decreased by approximately 5 log10 CFU per g. Free phages were detected in all treatments and transductants were detected and confirmed by PCR in the silo containing wilted grass, pretreated with slurry and ensiled without formic acid. Conclusions Verocytotoxigenic E. coli may survive the ensiling process and the conditions encountered are sufficient to induce vtx2 bacteriophage leading to low levels of phage-mediated vtx2 gene transfer. Significance and impact of the study These studies suggest that the ensiling of grass may create an environment which facilitates the emergence of new verocytotoxigenic E. coli.