Aquaculture is a fast growth activity in the world, but needs continued improvement in the conduct of management. In this sense, we highlight the Amazon basin, which has important species of fish as Colossoma macropomum with incipient production to meet the demand of the consumer market. Since that problem identified, we applied a Markovian decision process aiming to develop an optimization system to maximize the yield of the C. macropomum aquaculture. Were used mathematical algorithms were simulated with layout scenarios with 5 and 10 ponds, representing different size aquaculture farms. The transition between the growth phases was considered a stochastic process to satisfy the Markov property as per a sequential queuing through growth phases. The main goal was to define the target weight mix for the market and their optimal levels that optimize the production. The highest profitabilities were US$9,608 and 15,385 for 5 and 10 ponds layout scenarios, respectively, with a target weight mix harvest of 0.5 kg and 1 kg; 0.5 kg, 1 kg and 2 kg, respectively. The results showed the number of months for discounted the fixed monthly cost, about the cycle time that lasted of 5, 7 and 11 months, as well as the optimization was defined with the time at the fingerlings stage fixed for 50 days, possible to be improved.