This paper examines whether the risk attribution process is consistent with portfolio optimizations under tracking-error constraints and presents an extension of our previous work on risk attribution to other portfolio optimization contexts. It shows that optimization under the tracking-error constraint alone is consistent with the risk attribution process. Indeed, as soon as additional constraints (e.g. on total risk) are introduced, the risk attribution method conflicts with the performance attribution process, preventing us from legitimizing all of the optimal decisions taken by a portfolio manager.