This thesis modifies the original Ohlson equity evaluation model (OM) to evaluate stock price, risk premium and arbitrage region by considering the problem of incomplete stock market and risk-averse investor’s arbitrage behavior. The original Ohlson model assumes stock market is complete and investors are risk-nature. In fact, stock markets are always incomplete and almost all investors are risk-averse; therefore, in investing stock market investors face some risks and risk premium is the incentives to attract them to engage in investment. This implies that original Ohlson model need to be a nonlinear form. Based on this consideration, we modify the original Ohlson model by utilizing the concepts of smooth transition and arbitrage underlain in the STARX model to construct a nonlinear Ohlson evaluation model (NLOM). The nonlinear part in the NLOM could be an appropriate proxy variable for the other information in OM, and regarded as a risk premium to compensate for the possibility the investors face. Moreover, from the NLOM investors can set up arbitrage region. Empirical results show that most of the sample stock prices display smooth transition within different regimes and reveal a Logistic-STARX type; the NLOM we construct has significantly better goodness-of-fit and forecasting performance then OM. Most importantly, investors can easily employ the lagged book value, abnormal earning and stock price to find out arbitrage regions.