This study describes the realtime monitoring system for water quality conditions in coastal waters. Some issues on the data qualify control and quality analysis are examined along with examples of erroneous data. Three different cases of database produced by the realtime monitoring system are presented and analyzed, namely 1) hypoxic condition, 2) over-saturated D.O. and 3) short-term variability of temperature and D.O. In utilizing the realtime database, D.O. prediction and warning models are developed based on autoregressive stochastic process. The model is very simple, yet, users in various levels from powerful and useful with its ability to send warning messages to users in varous levels from governmental administrative staff to local fisherman, and give them some allowances to cope with the situation.