Abstract: The refined evaluation of reservoir parameters in development stage of oilfield is an important content of remaining oil study，because accuracy of parameters for calculation directly influences the reservoir calculation， establishment of reservoir model，reservoir evaluation and prediction of remaining oil distribution in development stage，reservoir parameters are obtained from log treatment and interpretation In fact，interpretation evaluation is a process to obtain quantitative data from log curves. But traditional interpretation is mostly limited in individual relationship inside geological body. We recommend treatment and interpretation with multi-well log data and emphasize truthfulness of change relationship between geological data and log data. In this case，there are advantages of abundant data and rich information. Different type reservoirs have different geological characteristics and corresponding log respond. We propose that log interpretation model can be established by characterizing quantitatively its changes to geological characteristics of different reservoirs. Taking Shengtuo oilfield as an example， we studied refined evaluation on reservoir parameters in development stage of oilfield our study consists of four research contents，including pretreatment and standardization of log data, variation analysis of reservoir tetrad property parameters in different development periods，establishment of log interpretation model and calculation of reservoir parameters ，fuzzy evaluation of reservoir parameters. Based on a lot of studies ，we believe that interpretation according to rate can be improved with tetrad property analysis on reservoir parameters，establishment log models using production test data and neural method . Results suggest that use of interpretation results and fuzzy mathematics method could make more accurate characterization and the distribution pattern of reservoir can be revealed. The study results on Shengtuo oilfield have been applied in remaining oil development and obtained good effects.