This paper mainly studies the relationship between subjective well-being （SWB） and three types of leisure consumption （achievement activities, social activities, time-out activities）. The characteristics of achievement activities, are usually competitive, especially in the process of leisure, which can make people feel certain challenge. Its essence is a learning activity. The characteristics of social activities are to have a " circle of friends”. The purpose is making friends, especially making friends with the colleagues or peers. Its essence is social. Time-out activities take the characteristic of relaxing. Its essence is pure time consumption. This paper focuses on the topic from both theoretical and empirical perspectives. Usually, SWB is a good proxy for utility. The difference between them is that utility cannot be measured directly, but SWB is a direct measure. So utility function and SWB function have close relationship. In the theoretical part, we establish the relationship between leisure time consumption and SWB （happiness） function based on the utility function. We further construct a happiness function which includes leisure activities. The defined leisure activities are divided into achievement activities, social activities, time-out activities to simplify later analysis. In the empirical part, we set the relationship between leisure activities and utility to be linear. It is a latent variable model. At first, we estimate the model by ordered probit estimation. The data is used based on Chinese General Social Survey （CGSS, 2010）. This is a nationally representative data. The meaning of leisure for individuals who do not have a job is totally different from the meaning of leisure for individuals who have a job because the unemployed have 24 hours of leisure. So this paper focuses on the sample who are employed. We drop the observations with missing value and also who do not have a job. The remained sample is 5725.  One of main contributions in this paper is that we consider possible report error for the dependent variable SWB because it is reported by him/herself. The problem is the fact that the estimators are usually biased when dependent variable has report errors. So we develop the misclassification probability model by adding the misclassification errors to draw the influence of errors from discrete data acquisition. The estimation results are obtained through the method of maximum likelihood estimation （MLE）. While comparing to the traditional ordered Probit estimates, our empirical results show that it is necessary and cannot be ignorable to adjust the errors in misclassification errors. Five of six misclassification factors are statistically significant at five percent significance level. Our conclusions are that （1） all the leisure activities have significant effects on SWB; （2） achievement activities and time-out activities have the positive impacts on SWB, whereas social activities have the negative impact.  This interesting finding implies that the change degrees of happiness which is brought by different choice of leisure activities are quite different. The influences of some unreasonable leisure arrangement even bring negative effect for personal welfare. So we must promote the efficiency of limited leisure consumption by arranging it reasonably. In modern society, it is worth paying close attention to the effectiveness of the social activities. Social activities like that maintaining and expanding a " circle of friends” may be tiring and exhausting and usually be not a good way to increase personal happiness. Current rapid social change may reduce the utility of social activities. Therefore, individual’s happiness does not simply depend on more leisure, but depends on appropriate choices of leisure activities and reasonable distribution of the leisure time. Make all kinds of leisure activities serve our happiness effectively. Then we can realize the happy dream, and finally the Chinese dream can be realized.