The scientific design of flood-control works is based on an evaluation of the hydrologic factors basic to flood events, particularly how rainfall and snow runoff, soil conditions, and channel influences can combine to produce greater or lesser floods. For this purpose an analysis of the pertinent hydrologic data is needed. The methods of analysis adopted should conform as closely as possible to those already in use and must be adapted to the quality of the available information. Maximum floods in 8 basins in New York and Pennsylvania during the winter and nonwinter months were studied, a total of 21 floods. The most outstanding winter flood of record in the North Atlantic region was that of March 1936. Rainfall plus snow melt in the basins studied ranged between 3.04 and 6.87 inches, and associated volumes of direct runoff from 1.88 to 5.63 inches. Winter floods have a common characteristic in their relation to freezing temperature. The antecedent periods, representing a period of snow accumulation and frost penetration, are below freezing, and the flood itself is contemporaneous with a period of above-freezing temperatures, usually associated with rain, during which the previously accumulated snow is melted. A second common characteristic of major winter floods is their tendency to be associated with widespread causal meteorologic conditions. There was a more complete conversion of rainfall and snow melt into runoff during the winter storms studied than during the wettest nonwinter flood. Snow melt during winter floods ranged from 0.04 to 0.07 inch per degree-day above 32° F. The depth of mean areal rainfall produced by the nonwinter storms studied ranged from 3.05 to 4.96 inches. The maximum 24-hour quantity at single stations was 14 inches, which was measured during the storm of July 1935 in New York. The volume of direct runoff ranged between 1.39 and 3.41 inches. The portion of rainfall that was converted into runoff varied in accordance with the rate of antecedent base flow, expressed in second-feet per square mile, and emphasized the influence of antecedent conditions. The average volume of direct runoff during winter floods was 4.24 inches, and the average during nonwinter floods was 2.44 inches. The latter, however, were more concentrated as to time, tending to compensate for large volume of runoff in winter, so that the crest rates of direct runoff averaged 0.056 inches per hour during the winter and 0.051 inches during the nonwinter period.