A methodology has been developed for the accurate assessment of localised reactivity perturbations in a BWR lattice embedded in a larger multiplying system, based on a full-system, unperturbed calculation, and on perturbed calculations on reduced-geometry models with reflective boundary conditions (typically, reflected-assembly calculations). Reflective reduced-geometry calculations are to be followed by a fast transferability correction for making the results representative of what full system computations would have produced. In this way, one can avoid the problem of having insufficient accuracy in the results (in spite of extremely lengthy iterations), particularly for cases of small reactivity effects. Furthermore, the factorization of reactivity effect transferability, a key feature of the developed methodology, provides valuable insight into the different effects contributing to a particular integral transferability factor, along with a quantification of the relative importance of these effects for each individually considered case. The initial investment, needed for realizing the relatively low required computational effort involved in the postcorrection procedure, is to obtain a limited number of adjoint equation solutions defined for the reference state at full system level. Application results are reported for the numerical analysis of fuel pin removal reactivity effects in LWR-PROTEUS. The latter is a programme of integral experiments, employing essentially a central LWR test zone driven critical by surrounding driver and buffer regions