In this work, a methodology is proposed to fit a survival model to a response surface in the presence of blocks, in order that this methodology allows improving the estimation of parameters and prediction in models where the variable of interest is observed in of time. Also, was developed an adaptation of the classical correction methods for ties data to the proposed methodology. Theoretical development was carried out for the construction, estimation, and validation of assumptions, which was developed using the Cox proportional hazard model and the response surfaces methodology. To evaluate the performance of the methodology in comparison with other methodologies, real and simulated data were used. The results also show the fact that by combining the proportional hazards model with the response surfaces methodology, it is possible to identify the levels of the treatments that optimize the response variable. Finally, it is concluded that this methodology has the advantage of being able to include a local control (block) that allows reducing experimental error, improving efficiency by detecting minor differences between treatments, which allows making comparisons over the treatments more reliable.