Small and steep rivers responds very rapidly to rain and athe crest stage is reached in a matter of a few hours. This poses the problem of inadequate time for warning potential victims of an impending flood. It is therefore important to have a technique that give a reasonbly reliable forecast within a short time, as any time gained in forecast preparation can be made available for warning and arelief purpose. to conform to this requirement the community Self-help Method for flash flood was developed. The flash-food models is able to give reasonably accurate forecast within a short time and is adaptable to basin changes. it is the aim of this thesis to apply the model on Gombak Rive. The flash food model requires the preparation of a set of 6 flood warning tables form which forecasts are made. Each table is printed with probable crest stage values which depend on 4 variations namely (i) the soil moisture deficit (ii) the total rain since the start of teh storm (iii) the rainfall amount for the last time period (iv) the duration of the storm. The warning tables are prepared by means of computer model. The wrning tables for Gombak River have been obtained and the checks made on them showed the are reasonably accurate.