This paper has developed new conservatism measures. i.e .. C-scores (Ct and Ci), which can distinguish the relative persistence of conservatism effects on future earnings and return. by extending the theories embedded in prior conservatism measures. i.e .. LC (lag component). BC (bias component) and Qa, Qb, respectively. suggested by Beaver and Ryan (2000). and Penman and Zhang (2002). Although Beaver and Ryan has successfully decomposed BTM (book-to-market ratio) into temporary (LC) and persistent (BC) components. their decomposition model is not based on accounting numbers. and too statistical that both LC and BC cannot be descriptively interpreted as accounting phenomena. Unlike Beaver and Ryan. Penman and Zhang incorporate some accounting numbers (not accruals). incurred by conservative accounting. into their Q scores. but they have not tested the relative persistence of each Q score's earnings effects. mainly concentrating on its relationship to earnings quality. This is why this paper suggests Ct and Ci. as new conservatism measures. which are purely based on accruals. Ct and Ci respectively measure the inter-temporal and cross-sectional differences of five accrual numbers. i.e.. total. operating. non-operating. discretionary. and non-discretionary ones. They are expected to capture short- and long-term conservatism effects on future earnings and return. respectively. The paper reports following evidence. supporting the empirical validity of Ct and Ci, as the short- and long-run measures of accounting conservatism. each of which reflects its temporary and persistent earnings-return effects. respectively. Firstly, the correlation of Ct and Ci to prior conservatism measures. including BTM, Qa, Qb, LC, and BC shows that both C scores have significantly positive associations with BTM, as the overall measure of conservatism. Not Ci but Ct scores show positive relationships to Qa, a short-term conservatism measure, whereas only Ci scores do to long-term measures, i.e.. Qb and BC. Secondly. the result of future ROE regressions on the current ROE and C scores shows that most Ci scores have significantly positive coefficients for future ROE. The positive relationship between Ci and ROE persists up to five years. On the contrary, most Ct scores shows insignificant. and inconsistent coefficients for future ROE, except for their positive relationship to ROEt+r. Consistently with the regression result. the ROE difference analysis between high and low C portfolios reveals that high Ci portfolio has higher ROE than low Ci portfolio for all periods (t-5~t+5), while high and low Ct portfolios shows no clear ROE differences for all periods other than t-1~t+1. when they experience the short-term inter-temporal sign reversal in ROE differences. Thirdly, the future return regressions on the current ROE and C scores provide contradictory results between Ci and Ct scores. efs coefficients for both Rt+4 and Rt+5 are significantly positive, while their coefficients for short-run future returns (Rt+1~Rt+5) are insignificant. and inconsistent in signs. Contrarily. C scores just show insignificant. but consistently positive coefficients for short-run returns, i.e., Rt+1 and Rt+2. and inconsistent coefficients for long-run returns. The additional return regressions. applying the three factor model of Fama and French (1995), show similar results. As the fourth explanatory variable. both Ci and Ct scores have significantly positive associations. respectively. with long-run (Rt+4 and Rt+5) and short-run (Rt+1) returns. Analyses of size-adjusted return (SAR) differences between high and low C portfolios render more apparent evidence on the relative persistence of return effects. which Ci reflects differently from Ct. The SAR difference between high and low Ci portfolios changes into positive at t+3. t+4. and t+5 from negative at t. while the positive BAR difference for Ct portfolios just occurs at t-l and t. Finally. to identify the underlying factors which enable Ci to be the long-term conservatism measure. capturing its persistent earnings and return effects. I regress Ci scores on five risk factors. and separately on nine growth factors. Among five risk factors. market beta. firm size, and BTM show significantly positive relationships to all Ci scores. Also. dividend payout ratio. earnings retention ratio. ratio of intangibles to total assets. and gro th in total asse s. among nine growth fac ors. have s rang positive associ a ions with Ci scores. These results sugges that Ci, as a long-term conservatism measure, reflects a firm's risk and growth related properties. which accounting earnings under conserva ism convey to the market. C scores could be applied to various research areas in accounting and finance. For example. C scores can control the accounting conservatism's effects on the book value in BTM anomaly studies. so that researchers can focus solely on the market's reaction to unbiased book values, C scores. as new accounting risk measures, might also be effectively used to improve the empirical validity of return-risk models. e.g.. Fama and French's three factor model. by increasing their explanatory power, and with their better specification. Further researches on C scores. which examine more detailed differences in conservatism effects among those five accruals, could find that one C score's conservatism effects may differ from other C scores', For operating accruals, as an example. Ci is far more efficient in capturing conservatism effects than Ct. whereas Ct is most efficient for non-operating accruals.