Weather watch/warning systems have been established for human health outcomes. Our study aims to develop and demonstrate a weather watch/warning system for asthma and stroke within the whole of South Korea, using a stratified regression approach. We converted claim-based health insurance data covering almost all medical claims for the only health insurance system in Korea for asthma and stroke from 1996-2003 into personalized disease episode data, and combined them with meteorological data. We utilized a step-wise regression method using factors extracted from the meteorological data to develop stratified models for six (stroke) and nine (asthma) regional and day-of-week strata. Validation studies showed that the actual number of hospitalizations in 2003 increased according to the three-leveled predictions (levels I, II, and III) from the model based on the 1996-2002 data. This system is accessible via the internet (http://industry.kma.go.kr/APP/sub_APP15_H01.htm) at the Korean Meteorological Administration website.