It is well known that usual ANOVA-F test becomes invalid if error variances are not equal. New test procedures are developed in this paper considering a (pxq) asymmetric factorial experiment when error variances vary from cell to cell. Such test procedures can be used for practical purposes for real field data.