Foodborne illnesses involving raw and minimally processed foods are often caused by human noroviruses (HuNoV) and hepatitis A virus (HAV). Since food is contaminated usually with small numbers of virions, these must be eluted from the food surface and then concentrated for detection. The objective of this study was to optimize an ultrafiltration (UF) concentration method for HAV and HuNoVs present on various fresh and frozen produce. The detection range of the optimized method and its applicability to different food matrices was compared to the reference method ISO 15216-1:2017. Strawberry, raspberry, blackberry, lettuce, and green onion (25 g) were contaminated with HAV, HuNoV GI.7 and HuNoV GII.4 and then recovered therefrom by elution. A commercial benchtop UF device was used for the concentration step. Viral RNA was extracted and detected by RT-qPCR. From fresh strawberries, recovery of HAV loaded at 104 genome copies per sample was 30 ± 13 %, elution time had no significant impact, and UF membrane with an 80–100 kDa cut-off in combination with Tris-glycine elution buffer at pH 9.5 was found optimal. At lower copy numbers on fresh strawberry, at least 1 log lower numbers of HuNoV were detectable by the UF method (103 vs 104 GII.4 copies/sample and 101 vs 103 GI.7 copies/sample), while HAV was detected at 101 genome copies/sample by both methods. Except on raspberry, the UF method was usually equivalent to the ISO method regardless of the virus tested. The UF method makes rapid viral concentration possible, while supporting the filtration of large volume of sample. With fewer steps and shorter analysis time than the ISO method, this method could be suitable for routine analysis of viruses throughout the food production and surveillance chain.