Most recent conditional asset pricing models are evaluated by the static Fama-MacBeth cross-sec-tional regressions，therefore the time-varying risk cannot be evaluated by constant risk loading and risk premi-ums． This paper，from the economic perspectives，applies a brand-new method—The dynamic cross-sectional regression—To investigate the performances of conditional asset pricing models: whether the time-varying re-turns can be explained by the time-varying risk premiums． Theoretically，this paper evidences that returns on assets depend on the linear risk premium function and innovations of the economy． Empirically，the paper tests the conditional asset pricing models’pricing performances based on Chinese and US stock markets． The paper finds that the short-term reversal rate and the turnover rate as the conditional variables can help CAPM and CCAPM to explain several test assets’time-varying returns． Moreover，this paper also tests the classic con-ditional asset pricing models in explaining different assets’time-varying returns． The paper finds that the persis-tent and slow-moving conditional variables can be better candidates for our conditional asset pricing models.