The observer performance for video monitor images presented in positive mode (‘bones black’) was compared to that of images presented in negative mode (‘bones white’) in a digital system for chest radiography based on a large image intensifier. In the first part of the study (I), the original, non-linear grey-scale reversal performed in the computer of the digital image intensifier system (DS 1000) was studied together with fixed and variable settings of light and contrast. In the second part of the study (II), true grey-scale reversal performed in a minicomputer (Micro VAX II) where the images were also processed with edge and contrast enhancement, was studied. The time spent viewing the images was also recorded. Before the second part of the study, the image intensifier system was optimized with regard to spatial and contrast resolution and dose settings. Simulated pathology was randomly positioned over the lungs and the mediastinum of an anthropomorphic phantom. Observer performance was evaluated with Receiver Operating Characteristic (ROC) analysis on the digital images. In the first part of the study, a significant advantage for the positive images was seen, but no significant difference between fixed and variable settings of light and contrast. In the second part of the study, no significant difference in the detectability of the test structures was seen between positive and negative images. A significant improvement in the detectability was seen between the first and the second part of the study for the nodules over the mediastinum in both negative and positive images. No significant difference in the time spent observing positive and negative images, was seen. If negative mode is chosen for image presentation, true grey-scale reversal is necessary for adequate contrast resolution.