Abstract. Rainfall during summer monsoon season (June–September; JJAS), that accounts for about 80 % of the yearly total over Indian region, is herald by its onset over Kerala in June. And these four summer monsoon months contribute 19 %, 32 %, 29 % and 20 % of the seasonal rainfall, respectively. Therefore, it is important that this seasonal cycle is captured by general circulation models (GCMs) used to understand and predict monsoon. In this study, using decade-long simulations of an atmospheric GCM, we show that surface hydrology over India as well as over its surrounding regions plays a central role during the onset phase of monsoon and thus modulates seasonal cycle. The model, in its default configuration, simulates early onset and excess precipitation (about double of that observed) over the Gangetic plain (GP) in June. Moreover, the model has large positive surface soil moisture bias over India throughout the year and negative bias over the arid-semiarid regions to the north-west of India during the pre-monsoon months. From multiple sensitivity experiments, it is discerned that the remote dry soil moisture bias in the model over the Western Central Asia region intensifies the tropospheric low-level circulation causing excessive moisture advection, followed by moisture convergence over the GP in the beginning of June, and an early onset. Local soil moisture over GP makes a diminutive contribution to precipitation bias in June. But as the season progresses and the remote influence weakens, the increased local soil moisture regulates surface and near-surface conditions which subsequently reduces moisture convergence over GP, reducing precipitation in the later phase of monsoon. The results presented here can be useful for diagnosis and improvement of land surface models.