Aggregate production planning (APP) in real systems often encounters uncertainties such as uncertain demand, production capacity with tolerances, and processing time. In a multi-period APP, processing time can change with work repetition due to the worker’s learning process. This worker’s learning process is often called the learning curve. APP modeling often considers product characteristics such as product defects. Quality improvements can be made to reduce defective products. This research aims to minimize total production costs in APP. Minimizing total production costs is done by considering several things, such as processing time uncertainty, learning curve, and reduction of defect rates. Processing time uncertainties can be resolved with fuzzy mathematical programming by relying on the decision maker’s expertise in determining the uncertainty range and then representing it as fuzzy numbers. In this research, the worker’s learning process influences processing time. Individual learning type is used in this research because it makes it easier to know the processing time of each worker. This research also considers improving product quality by balancing the investment made by the company and the defect rate in the production process. An APP model was developed in this research by considering processing time uncertainty using the Jimenez Method and learning curve and decreasing the defect rate using the Thornton Method. Optimization results were obtained using LINGO 18.0 software with a total production cost of $ 52,807.3. Therefore, the resulting model can solve APP problems with the objective function of minimizing total production costs.