We propose stochastic models for predicting and analysing the production of tilapia (Oreochromis niloticus), lettuce (Lactuca sativa), and cucumber (Cucumis sativus) cultivated in an aquaponic system. Fish and plants were cultivated in a shade house using 30, 60, and 90 fish/m3 employing an NFT system. Results from Monte Carlo simulation showed that higher yields of tilapia and cucumber, as well as larger plant sizes, were obtained by stocking at the highest density (90 fish/m3). At this density, with 95% confidence, yields of tilapia varied from 39.60 to 59.26 kg/m3, the final length of lettuce leaves varied from 13.53 to 28.5 cm, the final length of cucumber plant varied from 119 to 235.3 cm, and biomass of cucumber varied from 0.98 to 0.99 kg/m2. Regression and sensitivity analyses showed that dissolved oxygen, density, temperature, and electrical conductivity significantly affected the production of tilapia; density, nitrites, pH, and temperature influenced lettuce production; ammonium, pH, and density affected the production of cucumber plants; and ammonium and density influenced yields of cucumber. The greatest certainty to achieve higher yields of large tilapia was found at low densities. For plants, there was more certainty of harvesting larger products when cultivated with tilapia stocked at the highest density. A preliminary economic analysis of tilapia production showed that net revenues ranged from USD$ 18.50 to 81.76 per system, and that the best results were obtained when using the highest stocking density. We conclude that the models are useful for predicting and analysing the production of an aquaponic system.