A relatively inexpensive image analysis system has been developed to semi-automate the detection and quantification of microbial growth in sections of food. A system based on an IBM PC compatible, with a frame store card, was programmed to scan Gram-stained sections using a motorized stage. Each field of view was thresholded after subtraction of a background image and the area between two thresholds measured. In the food studied it was found that, by using a size limit, it was possible to reduce the number of fields that needed to be examined by a microscopist to approximately 3% of those scanned. Visual examination was still required to distinguish bacterial cells from other stained objects which occasionally occur.