For the India Meteorological Department (IMD), it has been an annual ritual for the past 20 years to issue its long-range monsoon forecast around this time of the year, though the credibility of these predictions has been on the wane. The projection that the rainfall during this year's monsoon season (June to September) will be equal to 99 per cent of the long-period average, should therefore be viewed with some caution, even if prima facie it is a good news for the country's farmers, economy and the hydrological balance. Last year, the IMD in its long-range forecast had reckoned the rainfall to be below normal by 6 per cent. In the event, it turned out to be above normal by the same margin. Since 1988, the IMD's prognosis hit the mark in only the first six years, when the 16-parameter power regression statistical monsoon prediction model (often called the Gowarikar model) proved accurate. Subsequent projections tended to be off the mark, leading to the discarding of this model in 2002, when it failed to forecast the drought that ensued. None of the statistical models evolved and tried out since then has been able to inspire much confidence in the accuracy of the monsoon prediction. The most significant modification in the monsoon prediction procedure was effected last year. Instead of a single mode, two new statistical ensemble systems were deployed