The experiences of the global financial crisis reveal that the spillover effects of the current global financial imbalances undermine the financial stability of different countries. In this emerging scenario, country-specific studies for identifying leading indicators of financial crisis appear important. This paper sets out an approach that develops an early warning model for banking crisis prediction in India, based on the ‘signals' approach and multivariate probit regression model. Initially, using the Index method for identifying monthly crisis dates, four episodes of systemic banking crisis have been found to have occurred during 1994-2007. Investigating these crisis episodes by the ‘signals' approach, results revealed increase in the spread between Bank Rate and 91-day T-Bill rate, increase in the ratio of short-term foreign debt to foreign exchange reserves, expansion of base money supply, economic slowdown, REER overvaluation and hike in LIBOR as some of the ‘leading' indicators for banking crisis. The study also finds that if the value of the weighted composite indicator, constructed with the ‘leading' indicators, exceeds 0.205, then the probability of banking crisis increases alarmingly in future months. Using probit regression, the identified indicators, determined by the ‘signals' approach, have been found to be robust and the early warning model reflects satisfactory performance, both in-sample and out-of-sample. The paper also identifies the relevant variables associated with low, medium and high states of banking fragility based on an ordered probit regression model.Overall, the results confirm the significance of global economic conditions working upon domestic macroeconomic variables and the sensitivity of the banking sector to crisis. Further, most of the identified indicators, in the study, started emitting sufficient early signals of an upcoming banking fragility, which could be indicative of necessary pre-emptive measures on the part of the authorities