This thesis studies the 1-minute information content of limit order books of individual stocks on the Taiwan Stock Exchange and employs it to predict one-minute-ahead stock returns. I find evidence that both the depth pattern of the limit order books and the best bid/ask-depth have prediction powers on stock prices, but the prediction power of the short-term historical price path is weak. The thesis also implements trading strategies and finds the profits based on the examined models disappear if transaction costs are considered, which supports that the weak-form efficient market hypothesis holds in the Taiwan stock market.