This work aims at improving the previously developed design procedure accounting for anisotropic dimensional changes on sintering. The goal of the project is both enlarging the reference database and investigating in depth the parameters responsible for anisotropic dimensional changes. The effects of green density, geometry, and compaction parameters (compaction speed hold down force and hold down time) on the anisotropy of dimensional changes on sintering were investigated. Ring shaped parts made of eight different iron-based materials were investigated in order to cover a large range of dimensional changes and different sintering mechanisms. The work is divided into two main parts. The first part focuses on investigating the effects of green density and geometry on the anisotropic dimensional changes, also enlarging the database. The application of the design methodology previously developed showed the significant role of anisotropy in the compaction plane for accurate prediction of dimensional changes, which was not highlighted previously. To solve this critical point a new anisotropy parameter is proposed and implemented in the design procedure. With the new anisotropy parameter, prediction of dimensional changes was improved while another critical point was highlighted, which is the scatter in the database. In order to have a reliable design procedure, database must be as accurate as possible, as demonstrated by a careful analysis on data processing. In the second part, the effect of compaction parameters on the anisotropy of dimensional changes on sintering was investigated, with a focus on the anisotropy in the compaction plane highlighted in the first part. No direct correlations were found between anisotropy of dimensional changes and compaction parameters, but compaction settings leading to lower anisotropy in the compaction plane were highlighted. Application of design methodology confirmed that the accuracy of predictions is higher when the sampling is less scattered. Also, a numerical study based on the experimental results was done to evaluate the possibility of neglecting anisotropy in the compaction plane for predicting dimensional changes. It was demonstrated that neglecting anisotropy in the compaction plane is acceptable, for some materials produced using the compaction settings minimizing such anisotropy. Additionally, a correlation between springback during ejection of the parts after compaction and anisotropy in the compaction plane was found.