The extreme active and break phases of the Indian Summer Monsoon (ISM) are sources of natural disasters such as flash floods, landslide or severe droughts. Thus, understanding and prediction of these extreme phases can have the greatest economic benefit. The primary interest of the present study is to examine the tropical convection and extratropical circulation conditions over a range of antecedent period associated with the extreme phases of active/break monsoon for the years 1979-2005. To distinguish the impact of tropical-extratropical interaction on the different location of the ISM, the whole domain is divided into north (NISM) and south parts (SISM). Based on an area-mean outgoing longwave radiation (OLR) index covering the NISM or SISM, the extreme phases that have absolute values of this index greater than 1.7 standard deviation over summer season (June-September) were selected for analysis. Composite maps of OLR and 200-hPa geopotential height preceding the extreme phases are produced to show the prominent leading systems in the tropics and extratropics. We found that the NISM extreme active (break) phases are immediately preceded by the concurrence of the upper tropospheric anomalous high (low) over Central Asia and enhanced (suppressed) elongated tropical convection zone collectively migrating from the Equatorial Indian Ocean (EIO) and the South China Sea to the NISM. An antecedent stationary Rossby wave train propagating from northwestern Europe to Central Asia via the western Siberia plain appears to induce developing of anomalous circulation over Central Asia. The phase relationship between building up of the anomalous circulation over Central Asia and propagation of tropical ISOs from the tropics to the NISM are critical for the likelihood of the occurrence of the NISM extreme phases. Without an appropriate timing of tropical-extratropical interaction, the active and break phases of the NISM hardly grow to an extreme condition. The antecedent ISOs originated from the EIO play a key role in determining the extreme phases of the SISM and the effect of extratropical circulation is trivial. To predict the NISM extreme phases, normalized 200-hPa geopotential height over the Central Asia (65šE-85šE, 32.5šN-42.5šN) and OLR (60šE-95šE, 5šN-15šN) to the south of the NISM are selected as two predictors. The combination of two predictors shows considerably greater skill than each predictor and explains almost 40% of the variance of the NISM OLR at lead time of 4-5 days. A simple prediction scenario is proposed using the mean of two predictors. Once it increases up to 1.0, an extreme phase is anticipated to occur over the NISM after 6-7 days. In general, an event forecasted by this simple scenario has a probability of 40% to develop to an extreme phase. This study suggests the importance of the tropical-extratropical interaction to the occurrence of the NISM extreme phases and the feasibility of forecasting the NISM extreme phases.