In view of several controversial EXAFS results on cobalamins and B-12 model
compounds published in the last ten years, a methodological study is prese
nted to explain the origin of these controversial results and to propose ho
w to improve EXAFS structural-parameter refinement. The advantages and limi
tations of first-shell filtering, constrained refinement, and the importanc
e of the mapping technique are described in order to suggest and encourage
improvement of data-analysis codes. The conclusions can be generalized and
applied to the EXAFS analysis of the majority of metal complexes.