A study was carried out to analyze the rainfall data for crop planning in rainfed regions,, Probability analysis of rainfall offers a better scope for predicting the minimum assured rainfall to help in crop planning. The 75% probability of rainfall occurrence was proposed as an index of dependable rainfall for crop production. Considering this in view, 25 years (1981-2005) of rainfall data of Mymensingh district have been analyzed in greater details covering probability aspects. Assured rainfall analysis, probability of potential evapotranspiration, water-balance approach and actual evapotranspiration were found quite effective to assess the water availability period for crop planning under rainfed condition. It was found that maximum rainfall occurred in July and minimum in December. The numbers of rainy days per year at 75% probability were 128 days. In the month of July and August numbers of rainy days were 20 and in December, it reduced to zero. The highest potential evapotranspiration was obtained in April (4.6 mm/day) and the lowest in January (2.4 mm/day).It was observed that April to October is the rainfall excess (rainfall>PET) and November to March is the deficit period (PET>rainfall). Climatic water balance revealed the possibility of water harvesting during May to October. It may be concluded that probabilistic rainfall data and number of rainy days, would be quite helpful for crop planning. It provides useful information like land preparation, planting, transplanting, intercultural operations, harvesting, threshing, drying and other cropping practices. This would go a long way in crop planning at farmers and at policy making level.