Several fallout patterns for Castle Bravo were prepared in tne late 1950's. Some of the better known patterns apoear in Ret. | and were preoarad by(4) the Air Force Special Weapons Project, (b) the Naval Radiological Defense Laooratory, and (ec) the Rand Corporation. A comparison of these three patterns shows significant differences in the maximum dose rates, as well as the shapes of the contours. This is due ia large part to the subjectivity involved in the calculations. Portions of the AFSWP and NRDL contours were based on dose rate measurements at Rongelap, Rongerik, and Utirik, as well as a crude estimate of the dose rate received by the Japanese fishing ship, the Lucky Dragon. The remainder of these patterns were obtained using the observed winds in a subjective manner to bend the pattern and achieve an approximate mass balance. The Rand contours used estimated winds between Bikini and Rongelap. These winds were obtained from interpolation of streamline analyses at several levels at different times. By contrast, the altered versions of the MATHEW-ADPIC codes used in this report allow us to use the observed winds at different locations and different times after detonation. No artificial bending of the pattern is required. The only subjectivity lies in the selection of code input parameters. At all times, the codes automatically assure conservation of mass. COMPUTER CODES The suite of codes developed for the Atmospheric Release Advisory Capability (ARAC) were extensively modified in order to incorporate a larger number of upper air wind levels. All prior uses of the codes have been to handle calculations for releases that did not rise higher than a few kilometers. Also, the standard ARAC codes do not involve sophisticated gravitational fall velocity calculations, nor do they include time-integrated deoosition.