Summary: Investors are assumed to be rational, however, empirical evidence shows otherwise. Investors categorize their investments into different mental accounts (MAs) with different biases. Behavioral portfolio theory (BPT) takes these behaviors and MAs into account when selecting for optimal portfolios. This study presents an aggregated portfolio optimization procedure using the framework of BPT. The procedure consists of three parts: return estimation, return weighting and MAs selection. Returns are estimated considering indexes that may reflect investor biases; Return estimates are then weighted according to SP/A theory to reflect investors’ perception; Portfolios are then selected using an integrated optimization model where both the safety-first and risk-seeking MAs are considered. The estimation and weighting parameters used are based on market index forecasts. The back-test results show that the resulting aggregate portfolio and the embedded portfolios of the two MAs can outperform the market and mean-variance portfolio.