High content screening (HCS) is becoming widely adopted as a high throughput screening modality, using hundred-of-thousands compounds library. The use of machine learning and artificial intelligence in image analysis is amplifying this trend. Another factor is the recognition that diverse cell phenotypes can be associated with changes in biological pathways relevant to disease processes. There are numerous challenges in HCS campaigns. These include limited ability to support replicates, low availability of precious and unique cells or reagents, high number of experimental batches, lengthy preparation of cells for imaging, image acquisition time (45–60 min per plate) and image processing time, deterioration of image quality with time post cell fixation and variability within wells and batches. To take advantage of the data in HCS, cell population based rather than well-based analyses are required. Historically, statistical analysis and hypothesis testing played only a limited role in non-high content high throughput campaigns. Thus, only a limited number of standard statistical criteria for hit selection in HCS have been developed so far. In addition to complex biological content in HCS campaigns, additional variability is impacted by cell and reagent handling and by instruments which may malfunction or perform unevenly. Together these can cause a significant number of wells or plates to fail. Here we describe an automated approach for hit analysis and detection in HCS. Our approach automates HCS hit detection using a methodology that is based on a documented statistical framework. We introduce the Virtual Plate concept in which selected wells from different plates are collated into a new, virtual plate. This allows the rescue and analysis of compound wells which have failed due to technical issues as well as to collect hit wells into one plate, allowing the user easier access to the hit data.