Intra-lot, inter-scan and other variabilities in radiochromic film dosimetry may have a severe impact on absolute dosimetry with this dosimeter. In the literature, several dosimetry protocols may be found characterized by different calibration functions and different film response variables. Also, the re-calibration methods found in the literature correct and minimize the impact of the variabilities in the absolute dose estimates. In this work, several recalibration methods and dosimetry protocols are evaluated. In order to find optimal configurations, their accuracy is compared, and the accuracy level that can be reached in each case is discussed. The efficient protocol and the parameter escalation are used to recalibrate EBT3 films from two different film batches. The mean absolute deviations between known doses and estimated doses for eight dose levels are obtained and compared with the self calibration of each reading, named intrinsic film calibration. Eight film sheets from two different lots and two digitizers are used. The parameter escalation method with a four-level recalibration using net optical density (NOD) and a power law as dosimetry protocol obtains the highest accuracy. Regarding the number of control strips, increasing the number from two to three makes the parameter escalation protocol to come close to intrinsic film calibration in all cases, but has a less important effect on the efficient protocol. Regardless the choice of the sensitometric variables, using the appropriate recalibration method results in accuracy levels typical of self calibration of the film. In addition, the parameter escalation method provides better results than the efficient protocol with three calibration strips.