Storing or transmitting images of small size cause no problem -when the storage capability is small or the channel bandwidth is not wide. Although the channels have been increased in bandwidth and several improvements were made to increase the storage capability, the size of image data continues to be larger and larger (For Landsat imagery 6xl09 bits are required to be stored every day). Also, the number of users of personal computers (PC) has increased which results in tremendous amounts of data being transferred between computers and terminals. The solution to these problems is that the images must be compressed. In numerous applications error free compression is the only acceptable means of data reduction One such application is the archival of medical or business documents, where lossy compression usually is prohibited for legal reasons. Another is digital radiography, where the loss of information can compromise diagnostic accuracy.In the present work a novel error free Experimental system for iMAge compression based on Learning Automata Schemes, known as^EXMALAS, has been introduced and implemented. The principle operation of the system is the updating of the probabilities of the gray levels which are coded using Huffman code. Instead of updating a single probability the present system updates a set of probabilities called a Block, where the probabilities inside the block are unequally rewarded. Three schemes of learning automata have been simulated and promising results were obtained. A compressing ratio of more than 3:1 has been achieved. The results are compared with the entropy of the image and the average bit rate obtained when Huffman code is applied to the probabilities of the gray levels of the image.