A Quantitative Microbial Risk Assessment (QMRA) was performed to estimate the expected number of listeriosis cases due to the consumption, on the last day of shelf life, of 20 000 servings of multi-ingredient sandwiches produced by a medium scale food producer in Italy, by different population strata, defined by infection susceptibility (healthy, susceptible, transplant recipients and total population). First, all the sandwich ingredients were analysed for pH, Aw, salt and organic acids content and submitted to challenge tests at three different temperatures (4, 6 and 10 °C) to evaluate their suitability for L. monocytogenes growth. Next, a stochastic model was constructed simulating the contamination of the ingredients that were the best (bean cream) and worst (cheese cream) growth substrates. For each substrate, an exposure assessment was performed, estimating the number of L. monocytogenes within each serving. Then, two dose-response models were alternatively applied: the first used a fixed r value for each of the three population groups, while the second considered a variable r value (lognormal distribution), taking into account the variability in strain virulence and different host subpopulations susceptibility. The stochastic model predicted zero cases for total population for both the substrates by using the fixed r approach, while 3 cases were expected when a higher variability (in virulence and susceptibility) was considered in the model; the number of cases increased to 45–52 in the worst scenario (bean cream contamination) assuming all servings would be consumed by transplant recipients. An uncertainty analysis was performed by considering alternative scenarios: a higher mean bacterial concentration (+ 0.5 Log CFU/g) or higher standard deviation (+ 0.5) determined evident increases in the expected number of cases, almost doubling the risk. A similar effect was also exerted by an extended storage time (from 72 to 96 h), in particular in the worst case scenario. Finally, different protective interventions were evaluated (70/30 N₂/CO₂ packaging, home cooking or their combination). Both the interventions resulted in a strong decrease of the risk; MAP packaging, should be regarded as the most promising one, as it can be performed by the producer, who can assure a strict control of the treatment performances.