SummaryLinear regression equations are developed for predicting 10-day rainfall in each monsoon month (June through September) of 14 meteorological subdivisions lying mainly in Western, North-Western and Central India. The potential predictors considered are 5-day mean 700 mb level contour heights and their anomalies at 104 grid points covering an area between 10 ~ to 45~ and 20 ~ to 145~ and past 24 hour rainfall of the subdivision concerned. The 209 predictors are preliminarily screened by stagewise procedure and finally screened by forward selection procedure. The verification conducted on limited data of 12 cases revealed moderate success in June and September in predicting the 10-day rainfall of some subdivisions in 3 equally probable classes. The equations developed in the study are not suitable for operational use. Further work on this line is necessary before full potential of synoptic-statistical methods is realised. It is found that the contour heights over the Arabian sea and near Japan bear respectively significant negative and positive correlation coefficients with the 10-day rainfall of most of the subdivisions considered in all the months. Explanation of these relationships is attempted. Zusammenfassung Vorhersage zehntigiger Monmnniederschlfige fiber Indien