Fish is one of the foods that contain protein and has a high nutrient content, including minerals, vitamins, and unsaturated fats. The protein content in fish is relatively high. However, fish also contains adverse effects if it has undergone spoilage. Therefore, it is necessary to select fresh fish and non-fresh fish. Currently, fish freshness selection is still done manually by fishermen. This research builds a system that can identify freshness in mackerel which is expected to be an alternative solution in the process of selecting freshness of fish manually which takes a long time. By utilizing the TCS3200 sensor to identify the color of the fish and MQ135 to detect the smell of ammonia gas in the fish. This system uses Arduino Uno as the controller of the whole system. The results obtained from the selection process are in the form of fresh and not fresh fish that have been separated based on color and odor indicators, the data will be displayed on the LCD in the form of the accumulated weight of each fish that has been selected. The results of 30 times testing the entire system as much as 30 data obtained an accuracy value of 86.7%.