This paper mainly studied to building multivariate control charts of multi-dots alarm rules. For different multi-dots alarm rules, control limit parameters can be given by a kind of method of calculating average run length. Then the performances of those kinds of multivariate control schemes under different alarm rules were compared with Hotelling T2 chart, MCUSUM and MEWMA. We can find from this compare that those charts under different alarm rules have advantage in detecting small changes in the mean vector of a multivariate process. At last, an example is used to illustrate how this method can be used in practice.