Technical Abstract: Infrequent high precipitation events produce a disproportionally large amount of the annual surface runoff, soil erosion, nutrient movement, and watershed sediment yield. Numerical simulation of these watershed processes often lack sufficiently long weather data records to adequately capture the stochastic nature of high precipitation events needed to conduct long-term environmental impact assessment studies. In such cases, a suitable weather generator can produce the required precipitation data. However, two troublesome issues regarding the precipitation model affect the quality of generated synthetic precipitation: the possibility of non-uniform distribution of random number (RN) for some climate types, and a discontinuous pattern in the daily average of generated precipitation. Improvements with regard to both issues were developed and implemented in synthetic weather generator SYNTOR. Impacts of these improvements on generated precipitation were assessed using precipitation records at two National Weather Service stations. The results showed that the improvements produced synthetic daily precipitation that more accurately represented historical precipitation characteristics.